December 2009 - YaST Development - openSUSE Mailing Lists

[yast-devel] redesign of webyast users module
by Martin Schmidkunz 12 Jan '10

12 Jan '10

Hi, please find some ideas for a redesign of webYaST`s "Users" module at: http://home.arcor.de/martin.schmidkunz/ (wasn`t sure if it is a good idea to store all the jquery stuff in my Export...) It is based on rlihm`s mock up (http://gagarin.suse.de/) The basic concept is to show only the most important table entry on a top level (e.g. user name). Therefore there are fewer elements on the screen which eases the use of the module and increases its self-descriptivness and its task orientation. On a second level which the user accesses after clicking on the user name the user can choose whether to "edit" or "delete" the entry. Again it is avoided to show non-task related information. Before deleting an item the user is prompted an "Are you sure?" pop up in order to increase error tolerance and prevent the user from unintended data loss. "Edit" shows the settings of the list entry. If more settings are available they are organized in tabs. The "Edit" dialog is left either via "Save" or "Cancel" New list entries can be added via "Add New Item" button on top the list. Concerning "basic concept" * "Save"/"Cancel"/"Add" do not work properly in the mock up * "Add" should be a wizard * User needs to click "Save" twice (once in the edit tab and then in the main window) I am not sure whether this is intuitive. What do you think about that? * list should be sortable but I am not caple of implementing this technically Concerning "Users" module * When adding a user currently there are 8-9 settings which needed to be adjusted and some of them seem quite obscure to me. So, what do you think: which settings should be really editable via GUI? Currently there are: Full Name Groups Default Group (GID) Home Directory Do not create home directory Login Shell Login Name UID Number Password Full Name, Login Name, Password and Groups seem quite clear to me but is it really necessary to set login shell and UID in the GUI? And home directory? Wouldn`t it make sense to have a sane default which could be edited via an editor? * What do you think about showing permissions in the users module as well? This would seem quite logical to me. What is the reason, why permissions and users are separated? Do you have any other topics concerning users module GUI? Looking forward to your comments! Best regards and have a happy 2010, Martin ---------------------------------------------------------------- Martin Schmidkunz User Experience Specialist martin.schmidkunz(a)novell.com +49 (0) 911 740 53-346 ----------------------------------------------------------------- SUSE LINUX Products GmbH, GF: Markus Rex, HRB 16746 (AG Nürnberg) ----------------------------------------------------------------- Novell, Inc. Novell® Making IT Work As One™

3 6

[yast-devel] Printer
by Donn Washburn 12 Jan '10

12 Jan '10

Hey Group; This may be the wrong place to bring this up. However, it is a perceived Yast2 Printer problem. I have a Bother MFC-7840W USB laser printer that according to www.linuxprinter.org it works with Brothers drivers - I have them installed. lsusb and hwinfo --usb find it. And modprobe has all the modules required. But Yast2>Hardware>Printer under Add connection wizard USB will not show any devices or my USB printer. Why would this be the case. I went to USB and as I remember I hit enter and on the right side a new window shows up and is always empty. Yast2 then will not advance until something is selected -- 73 de Donn Washburn 307 Savoy Street Email:" n5xwb(a)comcast.net " Sugar Land, TX 77478 LL# 1.281.242.3256 Ham Callsign N5XWB HAMs : " n5xwb(a)arrl.net " VoIP via Gizmo: bmw_87kbike / via Skype: n5xwbg BMW MOA #: 4146 - Ambassador " http://counter.li.org " #279316 -- To unsubscribe, e-mail: yast-devel+unsubscribe(a)opensuse.org For additional commands, e-mail: yast-devel+help(a)opensuse.org

3 4

[yast-devel] Webyast: HttpMock and utility for it
by Josef Reidinger 04 Jan '10

04 Jan '10

Hi, in short: in branch for yast model I add also injector to HttpMock (standard library which is part of ActiveResource to test Rest-service based models) so if you plan write test look at it (or wait until it is merged). in long: Main goal to add models to webclient instead of dynamic one is easier testing of functionality which should go to model as controller is only thin layer which serves data. But I found that it is not so easy due to our design and also not so simple situation with rest-service site data. From my point of view correct testing of rest-service is provide same data as expected from rest-service and test if it work as expected. Now if test is ever present (not much models, I already fill bug) then it is huge mock which in real doesn't test much functionality. So we need some mock service which mock only target rest-service. Why I choose HttpMock? Even if our local consultant suggest use another framework I choose http because it looks like sufficient for our goal and it is already in ActiveResource so we don't need another gems. Why not use HttpMock without utilities? Of course it is possible, but you must simulate resource which translate interface to path and also permissions service which is out of test focus and test should not contain permissions service details. Also there is problem with authentication which must be solved on your own. So I create injector which add useful methods to HttpMock. How to use it in test: 1) define response (we separate at least visually tested data and tests): TEST_RESPONSE = <<EOF <test> <arg1>test</arg1> </test> EOF 2) now mock network service ( I suggest use it in setup - but of course for testing exceptions you should redefine it in method) def setup ActiveResource::HttpMock.set_authentification #a) ActiveResource::HttpMock.respond_to do |mock| header = ActiveResource::HttpMock.authentification_header #b) mock.resources :'org.opensuse.yast.modules.test' => "/test", :'org.opensuse.yast.modules.test2' => "/test2" #c) mock.permissions "org.opensuse.yast.modules.test", { :synchronize => true } #d) mock.get "/test.xml", header, TEST_RESPONSE, 200 mock.post "/test.xml", header, TEST_RESPONSE, 200 mock.get "/test2.xml", header, TEST2_RESPONSE, 200 mock.post "/test2.xml", header, TEST2_RESPONSE, 200 end end a) this utility sets authentification and target machine to predefined values b) authentification_header get hash which is passed to authenticate. It is required in mock header. otherwise it is not matched, but it is hash, so you can add your own values. c) define resources ( often it is just one, but for example if you use two rest-service (as time use time and ntp) then you define both. after hash argument you pass hash with options, which contains key singleton and policy ( mentioned in previous mail why I found it useless so it is optional, without it I use default which is no policy and single resource) d) mock permissions controller, just pass prefix and then hash with permission and if it is granted rest of method is common HttpMock which return expected values, so you can freely test it. 3) test model or controller :) def test_find_and_save test = TestModel.find :one assert_equal "test",test.arg1 #see XML for arg1 value assert test.save end Note: This utility class can be used without yast_model base model, because also dynamic created ActiveResource use this resources (but I do not test it, so expect problems) If you don't have any complains or suggestion how to improve it I plan to merge it together with yast_model. welcome any questions, suggestions or complains Josef -- Josef Reidinger YaST team maintainer of perl-Bootloader, YaST2-Repair, webyast (language,time,basesystem,ntp) -- To unsubscribe, e-mail: yast-devel+unsubscribe(a)opensuse.org For additional commands, e-mail: yast-devel+help(a)opensuse.org

2 2

[yast-devel] redesign of webyast users module
by stefan.eisenwiener＠freenet.de 30 Dec '09

30 Dec '09

Hi, I found an EMail by Martin Schmidkunz about this in my Mailbox. Which made me look at the concept of YaST at a hole. YaST today has three (frontend) servers (ncurses, qt, gtk), which should do the garphical visualization of the YaST Modules. Dose Modules should comuicate through SCR with the system level. Now WebYaST was added, but not as an (frontend) server to YaST, but as an outside wrapper witch calls the cli of some Modules (dose with have one), and brings own Modules these Modules. My question now is why did you through a way the original fronted concept of YaST? Best regards Stefan Gesendet von freenetMail- Mehr als nur eine E-Mail-Adresse http://email.freenet.de/dienste/emailoffice/produktuebersicht/basic/mail/in…

2 1

[yast-devel] [ANNOUNCE] WebYaST now in gitorious.org/opensuse
by Duncan Mac-Vicar Prett 23 Dec '09

23 Dec '09

WebYaST is now available on http://gitorious.org/opensuse http://gitorious.org/opensuse/yast-rest-service http://gitorious.org/opensuse/yast-web-client http://gitorious.org/opensuse/ruby-polkit There is an initial opensuse-yast-developers team defined with access to those repositories. To switch your current checkout you only need to edit your .git/config origin remote If before it was git@git.opensuse.org:projects/yast/web-client.git Now it would be: git@gitorious.org:opensuse/yast-web-client.git To start, create your account at gitorious.org (you can use openID as well) and request to be added to opensuse-yast-developers group. (the best would be to have someone with admin permissions on each location). You have to configure your ssh public keys in your account settings then: http://gitorious.org/~yourlogin The repositories in git.opensuse.org are still accessible, but you can't push to them anymore. They will stay there until no repositories are left on that machine. gitorious provides feeds for each repository pushes, and per each branch commits with integrated commits reviews and merge requests. -- Duncan Mac-Vicar P. - Engineering Manager, YaST SUSE LINUX Products GmbH, GF: Markus Rex, HRB 16746 (AG Nuernberg) -- To unsubscribe, e-mail: yast-devel+unsubscribe(a)opensuse.org For additional commands, e-mail: yast-devel+help(a)opensuse.org

1 0

[yast-devel] Re: [yast-commit] <rest-service> master : cleanup code; bring metrix to the level of status
by Josef Reidinger 23 Dec '09

23 Dec '09

Stefan Schubert write: > ref: refs/heads/master > commit 4e4d3e055a49b1cf76c2082873b9674b2d8f13eb > Author: Stefan Schubert <schubi(a)suse.de> > Date: Tue Dec 22 11:03:14 2009 +0100 > > cleanup code; bring metrix to the level of status > --- > .../status/app/controllers/metrics_controller.rb | 1 - > plugins/status/app/models/metric.rb | 57 +++++++++++++++++++- > 2 files changed, 56 insertions(+), 2 deletions(-) > > diff --git a/plugins/status/app/controllers/metrics_controller.rb b/plugins/status/app/controllers/metrics_controller.rb > index c699898..d39205f 100644 > --- a/plugins/status/app/controllers/metrics_controller.rb > +++ b/plugins/status/app/controllers/metrics_controller.rb > @@ -1,6 +1,5 @@ > include ApplicationHelper > > -require 'scr' > require 'metric' > require 'uri' > > diff --git a/plugins/status/app/models/metric.rb b/plugins/status/app/models/metric.rb > index f3e0919..ef8ec02 100644 > --- a/plugins/status/app/models/metric.rb > +++ b/plugins/status/app/models/metric.rb > @@ -5,7 +5,26 @@ > # @author Duncan Mac-Vicar P. <dmacvicar(a)suse.de> > # @author Stefan Schubert <schubi(a)suse.de> > # > -require 'yaml' > +require 'exceptions' > + > +class CollectdOutOfSyncError < BackendException Hi, few notes :) I know that I also do it, but when you clean code I suggest to move this exception to lib directory of plugin (so plugins/status/lib/collectd_out_of_sync_error.rb to work rails autoload). > + def initialize(timestamp) > + @timestamp = timestamp > + super("Collectd is out of sync.") > + end > + > + def to_xml > + xml = Builder::XmlMarkup.new({}) > + xml.instruct! > + > + xml.error do > + xml.type "COLLECTD_SYNC_ERROR" > + xml.description "Collectd is out of sync. Status information can be expected at #{Time.at((a)timestamp.to_i).ctime}." > + > + xml.timestamp @timestamp > + end > + end > +end > > # > # This class acts as a model for metrics provided by collectd > @@ -30,6 +49,7 @@ require 'yaml' > # end > # > class Metric > + require 'yaml' > > COLLECTD_BASE = "/var/lib/collectd" > > @@ -41,6 +61,7 @@ class Metric > # convenience, plugin and instance > attr_reader :plugin_full > attr_reader :type_full > + attr_reader :config > > # like identifier, but to be used as a REST id > # so / are replaced by + > @@ -123,6 +144,18 @@ class Metric > # these values can be extracted from the file but we cache > # them > @host, @plugin, @plugin_instance, @type, @type_instance = Metric.parse_file_path(path) > + > + #find the correct plugin path for the config file > + plugin_config_dir = "#{RAILS_ROOT}/config" #default > + Rails.configuration.plugin_paths.each do |plugin_path| > + if File.directory?(File.join(plugin_path, "status")) > + plugin_config_dir = plugin_path+"/status/config" > + Dir.mkdir(plugin_config_dir) unless File.directory?(plugin_config_dir) ^^^ Attention: This doesn't work! yastws doesn't have write permissions to itself so you cannot create directory below RAILS_ROOT! (I already have same problem with writing css cache) > + break > + end > + end > + #reading configuration file > + @conf = YAML.load(File.open(File.join(plugin_config_dir, "status_configuration.yaml"))) if File.exists?(File.join(plugin_config_dir, "status_configuration.yaml")) ^^^^ I suggest store to temp variable File.join which you repeat. It is not DRY, effective and you have unnecessary long line: conf_file = File.join(plugin_config_dir, "status_configuration.yaml") @conf = YAML.load(File.open(conf_file)) if File.exists? conf_file > end > > # parses the host, plugin, plugin_instance, type and type_instance > @@ -226,6 +259,13 @@ class Metric > xml.type type > xml.type_instance type_instance > > + if @conf and @conf.has_key?(id) You can use try ( but I am not sure if it improve readability so use what you find better) if @conf.try(:has_key,id) > + xml.limits() do > + xml.max(@conf["#{id}"]["max"]) > + xml.min(@conf["#{id}"]["min"]) ^^^ I don't understand why you use "#{id}"...if you want string just use id.to_s ... if symbol, then use id.to_sym > + end > + end > + > # serialize data unless it is disabled > unless opts.has_key?(:data) and !opts[:data] > metric_data = data(data_opts) > @@ -242,6 +282,12 @@ class Metric > end > > end > + > + #get last entry of rrd database > + def self.last_db_entry(file) > + `/bin/sh -c "LC_ALL=C rrdtool last #{file}"` ^^^ I don't look at whole code, but you must be really sure, that file cannot be compromited otherwise it is security problem. Also be sure, then file doesn't contain special characters. so at least '#{file}' should be used as not much file has ' inside :) > + end > + > > # runs the rddtool on file with start time and end time > # > @@ -250,6 +296,15 @@ class Metric > # > # default last 5 minutes from now > def self.run_rrdtool(file, opts={}) > + > + #checking if collectd is running > + ServiceNotRunning.new('collectd') unless collectd_running? ^^^^ I think there is missing raise - test, test,test ;) if you change behavior, test it at first (test-driven development ;) > + > + #checking if systemtime is BEHIND the last entry of collectd. > + #If yes, no data will be available. > + timestamp = last_db_entry(file) > + raise CollectdOutOfSyncError.new(timestamp) if Time.now < Time.at(timestamp.to_i) > + > stop = opts.has_key?(:stop) ? opts[:stop] : Time.now > start = opts.has_key?(:start) ? opts[:start] : stop - 300 > > -- Josef Reidinger YaST team maintainer of perl-Bootloader, YaST2-Repair, webyast (language,time,basesystem,ntp) -- To unsubscribe, e-mail: yast-devel+unsubscribe(a)opensuse.org For additional commands, e-mail: yast-devel+help(a)opensuse.org

2 1

[yast-devel] vesele vanoce schoene weihnachten merry christmas
by Martin Schmidkunz 22 Dec '09

22 Dec '09

OK: time for christmas whishes :-) Preji Vam vsem klidne a spokojene vanoce, v novem roce spoutu stesti, zdravi a dobre pohody! Ich wünsch euch schöne, erholsame Weihnachten, einen guten Rutsch & alles Gute für 2010! I wish you happy and recreative Christmas, a Happy New Year with lots of health, joy and success! Best regards, Martin ---------------------------------------------------------------- Martin Schmidkunz User Experience Specialist martin.schmidkunz(a)novell.com +49 (0) 911 740 53-346 ----------------------------------------------------------------- SUSE LINUX Products GmbH, GF: Markus Rex, HRB 16746 (AG Nürnberg) ----------------------------------------------------------------- Novell, Inc. Novell® Making IT Work As One™

1 0

[yast-devel] Re: [yast-commit] <rest-service> master : code cleanup
by Josef Reidinger 22 Dec '09

22 Dec '09

Stefan Schubert write: > ref: refs/heads/master > commit 31f399a35839d17d62005b9623562d53af7a6092 > Author: Stefan Schubert <schubi(a)suse.de> > Date: Tue Dec 22 12:02:12 2009 +0100 > > code cleanup > --- > .../status/app/controllers/metrics_controller.rb | 36 ++++++------------- > 1 files changed, 12 insertions(+), 24 deletions(-) > > diff --git a/plugins/status/app/controllers/metrics_controller.rb b/plugins/status/app/controllers/metrics_controller.rb > index d39205f..719b6ee 100644 > --- a/plugins/status/app/controllers/metrics_controller.rb > +++ b/plugins/status/app/controllers/metrics_controller.rb > @@ -58,20 +58,10 @@ class MetricsController < ApplicationController > end > limits = Hash.new > limits = create_limit(params["status"]) > - f = File.open(File.join(plugin_config_dir, "status_limits.yaml"), "w") > + f = File.open(File.join(plugin_config_dir, "status_configuration.yaml"), "w") > f.write(limits.to_yaml) > f.close > - > - @status = Status.new > - # use now if time is not valid > - begin > - stop = params[:stop].blank? ? Time.now : Time.at(params[:stop].to_i) > - start = params[:start].blank? ? stop - 300 : Time.at(params[:start].to_i) > - @status.collect_data(start, stop, params[:data]) > - render :show > - rescue Exception => e > - render :text => e.to_s, :status => 400 # bad request > - end > + render :text => "OK" ^^^ I think that it is not good idea to render text, as you maybe want to check it on frontend and ActiveResource show have problem with it. > end > > # GET /status > @@ -88,12 +78,13 @@ class MetricsController < ApplicationController > end > end > > - @metrics = Metric.find(:all, conditions) > - > - respond_to do |format| > - format.xml { render :xml => @metrics.to_xml(:root => 'metrics', :data => false) } > - end > + @metric = Metric.find(:all, conditions) > > + @data = false > + @start = nil > + @stop = nil > + > + render :show ^^^^ I think you forget to add show template to render I see only status one. > end > > # GET /status/1 > @@ -102,14 +93,11 @@ class MetricsController < ApplicationController > #permission_check("org.opensuse.yast.system.status.read") > begin > @metric = Metric.find(params[:id]) > - stop = params[:stop].blank? ? Time.now : Time.at(params[:stop].to_i) > - start = params[:start].blank? ? stop - 300 : Time.at(params[:start].to_i) > - > - respond_to do |format| > - format.xml { render :xml => @metric.to_xml(:start => start, :stop => stop) } > - end > + @stop = params[:stop].blank? ? Time.now : Time.at(params[:stop].to_i) > + @start = params[:start].blank? ? @stop - 300 : Time.at(params[:start].to_i) ^^^^ 300 should be defined constant such as DEFAULT_RANGE=300 or similar which say context of number. > + @data = true > rescue Exception => e > - render :text => e.to_s, :status => 400 # bad request > + render ErrorResult.error(400, 108, e.to_s) and return ^^^ I think it should use same way for reporting exception as rest of webservice, then you don't need to handle client exception (and just server) on frontend. And if I understand code, you want to report bad request, which is typical due to bad parameters and for this purpose you should use InvalidParameters exception which correctly report it to ActiveResource, so you can easily check result of save method on frontend (it behaves as common rails application). > end > end > end > -- Josef Reidinger YaST team maintainer of perl-Bootloader, YaST2-Repair, webyast (language,time,basesystem,ntp) -- To unsubscribe, e-mail: yast-devel+unsubscribe(a)opensuse.org For additional commands, e-mail: yast-devel+help(a)opensuse.org

1 0

[yast-devel] Webyast: Models in webclient and some suggestion to rest-service resources
by Josef Reidinger 21 Dec '09

21 Dec '09

Hi, I just finish basic functionality of YastModel for webclient. I think that my solution is better then use_activeresource branch :) It is really intuitive to use. It is still only in branch waiting for your comments and improvements (branch jr-yastmodel). How use it? simply :) (example is time module in branch) create model: class Stime < ActiveResource::Base # 1) 2) extend YastModel::Base #3) model_interface :'org.opensuse.yast.modules.yapi.time' #4) end 1) Stime because Time already exist and we don't want to overwrite it 2) use ActiveResource not ActiveRecord (gives quite strange errors :) 3) contain definition of model_interface and needed fixtures 4) interface which you use now with load_proxy And model usage is then same as for common ActiveResource: @permissions = Stime.permissions #5) time = Stime.find :one #6) time.timezone etc... 5) It is class method because it is related to site, user and collection which is also stored in class, so I leave it in same scope 6) :one because time is single resource And there should be some discuss about format which provide resources. What is policy? Resource model in rest-service really lack any documentation Also I am not sure if we require singleton key, as type must know user at webclient and use find :all, :first:, :last etc for multi resource and :one for single resource. What I found inadequate is href, which should be separated to fixed prefix and changable collection name. Because now we have quite serius problems if we want use namespace for controllers. Feel free to post any question or suggestion to my solution. If martin agree with it I plan to merge it to master (as it is separate solution). Thanks Josef -- Josef Reidinger YaST team maintainer of perl-Bootloader, YaST2-Repair, webyast (language,time,basesystem,ntp) -- To unsubscribe, e-mail: yast-devel+unsubscribe(a)opensuse.org For additional commands, e-mail: yast-devel+help(a)opensuse.org

1 0

[yast-devel] Moving WebYaST bugs to standalone project
by Klaus Kaempf 21 Dec '09

21 Dec '09

Hi, we now have 'WebYaST' as a standalone project in bugzilla.novell.com, open to the public. Being open was the reason for staying under 'openSUSE'. Now I intend to move all bugs over to 'WebYaST'. The move will happen either later today or tomorrow. We have full control over components in Bugzilla, please mail me for suggestions. Klaus --- SUSE LINUX Products GmbH, GF: Markus Rex, HRB 16746 (AG Nürnberg) -- To unsubscribe, e-mail: yast-devel+unsubscribe(a)opensuse.org For additional commands, e-mail: yast-devel+help(a)opensuse.org

1 0

[yast-devel] HelpNeeded: ActiveResource and transitive inheritance
by Josef Reidinger 21 Dec '09

21 Dec '09

Hi, I create own wrapper over ActiveResource and found quite interesting problem which blocking my work. If I create ancestor of ActiveResource::Base then everything works, but when I create ancestor of ancestor then it stop working (exactly it return nil on method which should not return nil, because it is accessor with default and I assign value so it should return assigned value or default but not nil). I attach example code which demonstrate this problem. thanks for help Josef -- Josef Reidinger YaST team maintainer of perl-Bootloader, YaST2-Repair, webyast (language,time,basesystem,ntp)

1 0

[yast-devel] Re: [yast-commit] <web-client> master : handle PackageKit errors (bnc #559233)
by Josef Reidinger 18 Dec '09

18 Dec '09

Stefan Schubert write: > ref: refs/heads/master > commit 547aa5b85029bdfd4b37ef9464be2494508f4e0d > Author: Stefan Schubert <schubi(a)suse.de> > Date: Fri Dec 18 13:51:40 2009 +0100 > > handle PackageKit errors (bnc #559233) > --- > .../app/controllers/patch_updates_controller.rb | 32 ++++++++++++++++--- > .../views/patch_updates/_patch_summary.html.erb | 2 +- > .../app/views/patch_updates/index.html.erb | 2 +- > .../package/yast2-webclient-patch_updates.changes | 5 +++ > 4 files changed, 34 insertions(+), 7 deletions(-) > > diff --git a/plugins/patch_updates/app/controllers/patch_updates_controller.rb b/plugins/patch_updates/app/controllers/patch_updates_controller.rb > index 79d5f73..db14fb3 100644 > --- a/plugins/patch_updates/app/controllers/patch_updates_controller.rb > +++ b/plugins/patch_updates/app/controllers/patch_updates_controller.rb > @@ -13,7 +13,19 @@ class PatchUpdatesController < ApplicationController > # GET /patch_updates > # GET /patch_updates.xml > def index > - @patch_updates = load_proxy 'org.opensuse.yast.system.patches', :all > + begin > + @patch_updates = load_proxy 'org.opensuse.yast.system.patches', :all > + rescue ActiveResource::ServerError => e > + error_hash = Hash.from_xml e.response.body > + logger.warn error_hash.inspect > + if error_hash["error"] && error_hash["error"]["type"] == "PACKAGEKIT_ERROR" > + flash[:error] = error_hash["error"]["description"] > + @patch_updates = [] > + @error = true > + else > + raise e > + end > + end You can benefit from ClientException class which handles exceptions from backend. It contains implementation details which you can abstract ( and don't need to change code when implementation change). so your code could look like this one which looks more readable for me: rescue ActiveResource::ServerError => e ce = ClientException.new e if ce.backend_exception_type == "PACKAGEKIT_ERROR" flash[:error] = ce.message @patch_updates = [] @error = true else raise e end end > logger.debug "Available patches: #{(a)patch_updates.inspect}" > end > > @@ -23,11 +35,21 @@ class PatchUpdatesController < ApplicationController > patch_updates = nil > begin > patch_updates = load_proxy 'org.opensuse.yast.system.patches', :all > - rescue Exception => e > + rescue ActiveResource::ClientError => e > error = ClientException.new(e) > patch_updates = nil > - end > - > + error_string = _("A problem occured when loading patch information.") > + rescue ActiveResource::ServerError => e > + error_hash = Hash.from_xml e.response.body > + logger.warn error_hash.inspect > + if error_hash["error"] && error_hash["error"]["type"] == "PACKAGEKIT_ERROR" > + error_string = error_hash["error"]["description"] > + else > + error_string = _("A problem occured when loading patch information.") > + end > + error = ClientException.new(e) ^^^ I don't understand much why you set error, but you should benefit from it same as I show above, so: error_string = _("A problem occured when loading patch information.") error = ClientException.new(e) error_string = error.message if error.backend_exception_type == "PACKAGEKIT_ERROR" > + patch_updates = nil > + end > patches_summary = nil > <snip/> > -- Josef Reidinger YaST team maintainer of perl-Bootloader, YaST2-Repair, webyast (language,time,basesystem,ntp) -- To unsubscribe, e-mail: yast-devel+unsubscribe(a)opensuse.org For additional commands, e-mail: yast-devel+help(a)opensuse.org

2 1

[yast-devel] Y2DIR can contain multiple paths now
by Arvin Schnell 18 Dec '09

18 Dec '09

Hi, Y2DIR can contain multiple paths separated by ":" now. If you encounter any problems let me know. ciao Arvin -- Arvin Schnell, <aschnell(a)suse.de> Senior Software Engineer, Research & Development SUSE LINUX Products GmbH, GF: Markus Rex, HRB 16746 (AG Nürnberg) -- To unsubscribe, e-mail: yast-devel+unsubscribe(a)opensuse.org For additional commands, e-mail: yast-devel+help(a)opensuse.org

1 0

[yast-devel] Re: [yast-commit] <rest-service> master : handle PackageKit errors (bnc #559233)
by Josef Reidinger 18 Dec '09

18 Dec '09

Stefan Schubert write: > ref: refs/heads/master > commit d50ad4b891552d7f205325e408fca63ad53d5a89 > Author: Stefan Schubert <schubi(a)suse.de> > Date: Fri Dec 18 13:10:02 2009 +0100 > > handle PackageKit errors (bnc #559233) > --- > plugins/patches/app/models/resolvable.rb | 30 +++++++++++++++++++- > .../package/yast2-webservice-patches.changes | 5 +++ > 2 files changed, 34 insertions(+), 1 deletions(-) > > diff --git a/plugins/patches/app/models/resolvable.rb b/plugins/patches/app/models/resolvable.rb > index 09b1795..7b9acc3 100644 > --- a/plugins/patches/app/models/resolvable.rb > +++ b/plugins/patches/app/models/resolvable.rb > @@ -7,6 +7,23 @@ require 'thread' > > require 'exceptions' > > +class PackageKitError < BackendException > + def initialize(description) > + @description = description > + super("PackageKit error") > + end > + > + def to_xml > + xml = Builder::XmlMarkup.new({}) > + xml.instruct! > + > + xml.error do > + xml.type "PACKAGEKIT_ERROR" > + xml.description @description ^^^ Hi, duncan add to frontend ability to identify bugs. By default if exception doesn't have bug key it is considered as bug (duncan - it is good default?). So in this case it is not our bug, but it is problem of package kit, maybe we should not report it as bug - Duncan what is idea behind bug key and how it is used? thanks Josef > <snip/> -- Josef Reidinger YaST team maintainer of perl-Bootloader, YaST2-Repair, webyast (language,time,basesystem,ntp) -- To unsubscribe, e-mail: yast-devel+unsubscribe(a)opensuse.org For additional commands, e-mail: yast-devel+help(a)opensuse.org

2 1

[yast-devel] Re: [yast-commit] <rest-service> backgroud_patches_bnc550934 : testsuite update
by Josef Reidinger 18 Dec '09

18 Dec '09

Ladislav Slezak write: > ref: refs/heads/backgroud_patches_bnc550934 > commit cf8abb9c2f7d0b2a1ecf9c4e26e817cb00a67bef > Author: Ladislav Slezak <lslezak(a)novell.com> > Date: Fri Dec 18 10:04:23 2009 +0100 > > testsuite update > --- > webservice/test/unit/background_manager_test.rb | 4 ++++ > webservice/test/unit/background_status_test.rb | 2 +- > 2 files changed, 5 insertions(+), 1 deletions(-) > > diff --git a/webservice/test/unit/background_manager_test.rb b/webservice/test/unit/background_manager_test.rb > index 81f0b39..61a88cf 100644 > --- a/webservice/test/unit/background_manager_test.rb > +++ b/webservice/test/unit/background_manager_test.rb > @@ -20,10 +20,14 @@ class BackgroundManagerTest < ActiveSupport::TestCase > @bm.update_progress(:dummy) > assert_equal nil, @bm.get_progress(:dummy) > > + changed = false > @bm.update_progress(:dummy) do |s| > + # this block must NOT be executed > s.progress = 10 > + changed = true Hi, again some notes :) I personally don't like numbers and testing strings (even if I sometime especially in first iteration use it). I think that constant is much better. If you use the number or string only once it is OK, but when it is more times in code (as here, where you set it in init and then test if progress is same that change is not invoked) then I think constant is better. so something like INITIAL_PROGRESS = 10 s.progress = INITIAL_PROGRESS assert !changed > end > assert_equal nil, @bm.get_progress(:dummy) > + assert !changed > > > # register a process > diff --git a/webservice/test/unit/background_status_test.rb b/webservice/test/unit/background_status_test.rb > index 2131fa7..c63b93d 100644 > --- a/webservice/test/unit/background_status_test.rb > +++ b/webservice/test/unit/background_status_test.rb > @@ -38,7 +38,7 @@ class BackgroundStatusTest < ActiveSupport::TestCase > end > > def test_observing > - ot = ObserverTest.new(@bs, changed_flag) > + ot = ObserverTest.new(@bs) > s = 'dummy status' > p = 10 second part of first note > sp = 5 > If you use params then you close way to do backward compatible changes and also require strict order of parameters (see http://goruco2009.confreaks.com/30-may-2009-15-40-solid-object-oriented-des… ). So you can initialize by hash and then you can benefit from another module of BaseModel - MassAssignment which add method load, which takes hash and initialize variables by its values so you can then have: Class BackgroundStatus include BaseModel::MassAssignment def initialize(params={}) #some defaults set load params end if you use this, you can in future easy extend functionality just by adding new allowed key in hash and set default value for it. feel free to ask questions Josef -- Josef Reidinger YaST team maintainer of perl-Bootloader, YaST2-Repair, webyast (language,time,basesystem,ntp) -- To unsubscribe, e-mail: yast-devel+unsubscribe(a)opensuse.org For additional commands, e-mail: yast-devel+help(a)opensuse.org

1 0

[yast-devel] Opinions about WebYaST status page
by Stefan Schubert 17 Dec '09

17 Dec '09

Hi, I know the page is ugly :-) Thats why I am asking for opinions about new designs in a blog: http://lizards.opensuse.org/2009/12/15/2827/ If anyone has good ideas please add it to to block. Thank you ! Greetings Stefan P.S.: Heiko, Christian I have added you cause you have admin experiences and would like to ask you too. -- To unsubscribe, e-mail: yast-devel+unsubscribe(a)opensuse.org For additional commands, e-mail: yast-devel+help(a)opensuse.org

2 1

[yast-devel] Another presentation which is good to see
by Josef Reidinger 16 Dec '09

16 Dec '09

Hi, I see some videos and find two good which is good to see - both are from Scotland Ruby conference (and added to our learning source http://en.opensuse.org/YaST/Web/Development/TrainingKit/Learning_Ruby_and_R… ) http://scotland-on-rails.s3.amazonaws.com/2A04_DaveThomas-SOR.mp4 http://scotland-on-rails.s3.amazonaws.com/2A08_JoeOBrianJimWeirich-SOR.mp4 (whole list of scotland presentation is there ( I see other three videos, but I think it is not so useful) - http://www.rubyinside.com/scotland-on-rails-presentations-now-online-27-awe… ) The first one is about ruby object model, which describes how objects works in ruby and how you can benefit from it. It is good to see if you still has problem what means '<<' or 'class << self' or other tricks with objects and modules. The second one is about refactoring code and how to avoid common mistakes. It is imaginary dialog and it is quite funny (geek fun :). Good to see especially if you want use too much meta-programming after watching the first video (or how to write more robust test, how to improve bad test covered code etc.) Josef -- Josef Reidinger YaST team maintainer of perl-Bootloader, YaST2-Repair, webyast (language,time,basesystem,ntp) -- To unsubscribe, e-mail: yast-devel+unsubscribe(a)opensuse.org For additional commands, e-mail: yast-devel+help(a)opensuse.org

2 1

[yast-devel] Use ActiveResource Branch
by Josef Reidinger 16 Dec '09

16 Dec '09

Hi, I review this branch and I think with some improvement it should be merged to master. What should be changed: no init_and_find_method .. dynamic set information, but only once. My proposal: in set site get path from interface name and also set permissions (as part of site settings is set also user (see ARs sources for details)) overwrite site getter, that if super is nil, then set it from session add all necessary fixes of ARs and that is all what is needed to overwrite ( but some helpers could be add, so I suggest to add it as module ). I also recommend to not use Base for Permission and Resource resource, as it has fixed path and should use only ActiveResource (just with site setter before use). I think that after proposed change ActiveResource should be used as common ActiveResource only need interface setter: class Time < WebyastResource # I don't like current name has_interface :'org.opensuse.yast.modules.yapi.time' end If you agree I change implementation in branch and then merge it. -- Josef Reidinger YaST team maintainer of perl-Bootloader, YaST2-Repair, webyast (language,time,basesystem,ntp) -- To unsubscribe, e-mail: yast-devel+unsubscribe(a)opensuse.org For additional commands, e-mail: yast-devel+help(a)opensuse.org

1 0

[yast-devel] Re: [yast-commit] <rest-service> backgroud_patches_bnc550934 : patches and status - moved the shared background code
by Josef Reidinger 16 Dec '09

16 Dec '09

Ladislav Slezak write: > ref: refs/heads/backgroud_patches_bnc550934 > commit fe2894a7088810c100f1eac4a071a6911e8193ba > Author: Ladislav Slezak <lslezak(a)novell.com> > Date: Tue Dec 15 20:13:59 2009 +0100 > > patches and status - moved the shared background code > > to BackgroundManager module (for thread safe acess and better > readability) > --- > plugins/patches/app/models/patch.rb | 53 +++++++++------------- > plugins/status/lib/status.rb | 78 ++++++++++++++------------------ > webservice/lib/background_manager.rb | 80 ++++++++++++++++++++++++++++++++++ > 3 files changed, 135 insertions(+), 76 deletions(-) > > diff --git a/plugins/patches/app/models/patch.rb b/plugins/patches/app/models/patch.rb > index dd861de..a3608c6 100644 > --- a/plugins/patches/app/models/patch.rb > +++ b/plugins/patches/app/models/patch.rb > @@ -3,14 +3,10 @@ require 'resolvable' > # Model for patches available via package kit > class Patch < Resolvable > > - # class variables - they keep the information between requests > - # > - # currently running patch requests in background (current states) > - @@running = Hash.new > - # finished requests (actual results) > - @@done = Hash.new > - # a mutex which guards access to the shared class variables above > - @@mutex = Mutex.new > + # extend adds the module methods as class methods > + # (used instead of include because background processes > + # are used in class methods here) > + extend BackgroundManager > > def to_xml( options = {} ) > super :patch_update, options > @@ -35,7 +31,7 @@ class Patch < Resolvable > return patch_updates > end > > - def self.subprocess_find(what, bs) > + def self.subprocess_find(what) Hi, few notes. I think that name of variable is not good and quite confusing. Especially when you cal it below as e.g. "get_process what". I suggest to use variable name e.g. process_name or process_id or similar. > > # open subprocess > subproc = IO.popen(subprocess_command) > @@ -55,7 +51,8 @@ class Patch < Resolvable > result = received['patches'] > elsif received.has_key? 'background_status' > s = received['background_status'] > - @@mutex.synchronize do > + > + update_progress what do |bs| > bs.status = s['status'] > bs.progress = s['progress'] > bs.subprogress = s['subprogress'] > @@ -86,42 +83,34 @@ class Patch < Resolvable > > # background reading doesn't work correctly if class reloading is active > # (static class members are lost between requests) > - if background && !Rails.configuration.cache_classes > + if background && !background_enabled? > Rails.logger.info "Class reloading is active, cannot use background thread (set config.cache_classes = true)" > background = false > end > > if background > - # background status > - bs = nil > - @@mutex.synchronize do > - if @@done.has_key?(what) > - Rails.logger.debug "Request #{what} is done" > - return @@done.delete(what) > - end > - > - running = @@running[what] > - if running > - Rails.logger.debug "Request #{what} is already running: #{running.inspect}" > - return [running] > - end > + if process_finished? what > + Rails.logger.debug "Request #{what} is done" > + return get_value what > + end > > - bs = BackgroundStatus.new > - @@running[what] = bs > + running = get_progress what > + if running > + Rails.logger.debug "Request #{what} is already running: #{running.inspect}" > + return [running] > end > > + add_process what > + > Rails.logger.info "Starting background thread for reading patches..." > # run the patch query in a separate thread > Thread.new do > - res = subprocess_find(what, bs) > + res = subprocess_find what > Rails.logger.info "*** Patches thread: Found #{res.size} applicable patches" > - @@mutex.synchronize do > - @@running.delete(what) > - @@done[what] = res > - end > + finish_process(what, res) > end > > - return [bs] > + return [ get_progress(what) ] > else > return do_find(what) > end > diff --git a/plugins/status/lib/status.rb b/plugins/status/lib/status.rb > index 1bcdabe..cad3412 100644 > --- a/plugins/status/lib/status.rb > +++ b/plugins/status/lib/status.rb > @@ -10,12 +10,7 @@ class Status > :metrics, > :limits > > - # currently running status requests in background (current states) > - @@running = Hash.new > - # finished requests (actual results) > - @@done = Hash.new > - # a mutex which guards access to the shared class variables above > - @@mutex = Mutex.new > + include BackgroundManager > > def to_xml(options = {}) > if @data.class == Hash && @data.size == 1 && @data[:progress].class == BackgroundStatus > @@ -152,51 +147,48 @@ class Status > def draw_graph(heigth=nil, width=nil) > end > > + def create_unique_key(start, stop, data) > + start.to_i.to_s + '_' + stop.to_i.to_s + '_' + data.to_s > + end > + > # this is a wrapper to collect_data > def collect_data(start=nil, stop=nil, data = %w{cpu memory disk}, background = false) > # background reading doesn't work correctly if class reloading is active > # (static class members are lost between requests) > - if background && !Rails.configuration.cache_classes > + if background && !background_enabled? > Rails.logger.info "Class reloading is active, cannot use background thread (set config.cache_classes = true)" > background = false > end > > if background > - # background status > - bs = nil > - key = start.to_i.to_s + '_' + stop.to_i.to_s + '_' + data.to_s > - @@mutex.synchronize do > - if @@done.has_key?(key) > - Rails.logger.debug "Request #{key} is done" > - @data = @@done.delete key > - @data.delete :progress > - return @data > - end > + key = create_unique_key(start, stop, data) > > - running = @@running[key] > - if running > - Rails.logger.debug "Request #{key} is already running: #{running.inspect}" > - @data = {:progress => running} > - return @data > - end > + if process_finished? key > + Rails.logger.debug "Request #{key} is done" > + @data = get_value key > + @data.delete :progress > + return @data > + end > > - bs = BackgroundStatus.new > - @@running[key] = bs > + running = get_progress key > + if running > + Rails.logger.debug "Request #{key} is already running: #{running.inspect}" > + @data = {:progress => running} > + return @data > end > + > + add_process(key) > > # read the status in a separate thread > Thread.new do > Rails.logger.info '*** Background thread for reading status started...' > - res = collect_status_data(start, stop, data, bs) > - @@mutex.synchronize do > - @@running.delete(key) > - @@done[key] = res > - end > + res = collect_status_data(start, stop, data, true) > + finish_process(key, res) > Rails.logger.info '*** Finishing background status thread' > end > > # return current progress > - @data = {:progress => bs} > + @data = {:progress => get_progress(key)} > return @data > else > collect_status_data(start, stop, data) > @@ -204,8 +196,8 @@ class Status > end > > # creates several metrics for a defined period > - def collect_status_data(start=nil, stop=nil, data = %w{cpu memory disk}, progress=nil) > - > + def collect_status_data(start=nil, stop=nil, data = %w{cpu memory disk}, progress=false) > + key = create_unique_key(start, stop, data) if progress > metrics = available_metrics > #puts metrics.inspect > result = Hash.new > @@ -213,16 +205,14 @@ class Status > current_step = 0 if progress > case data > when nil, "all", "All" # all metrics > - if progress > - total_steps = metrics.size > - end > + total_steps = metrics.size if progress > > metrics.each_pair do |m, n| > if progress > - @@mutex.synchronize do > - progress.status = m.to_s > - progress.progress = 100 * current_step / total_steps > - Rails.logger.debug "*** Reading status: #{progress.status}: #{progress.progress}%" > + update_progress key do |stat| > + stat.status = m.to_s > + stat.progress = 100 * current_step / total_steps > + Rails.logger.debug "*** Reading status: #{stat.status}: #{stat.progress}%" > end > end > > @@ -238,10 +228,10 @@ class Status > total_steps = data.size if progress > data.each do |d| > if progress > - @@mutex.synchronize do > - progress.status = d.to_s > - progress.progress = 100 * current_step / total_steps > - Rails.logger.debug "*** Reading status: #{progress.status}: #{progress.progress}%" > + update_progress key do |stat| > + stat.status = d.to_s > + stat.progress = 100 * current_step / total_steps > + Rails.logger.debug "*** Reading status: #{stat.status}: #{stat.progress}%" > end > end > > diff --git a/webservice/lib/background_manager.rb b/webservice/lib/background_manager.rb > new file mode 100644 > index 0000000..722d078 > --- /dev/null > +++ b/webservice/lib/background_manager.rb > @@ -0,0 +1,80 @@ > + > +# this is a helper module for managing background processes with progress > +# for long running tasks > + > +module BackgroundManager > + # class variables - they keep the information between requests > + # DO NOT USE THEM DIRECTLY, USE THE THREAD SAFE METHODS BELOW! When I see what provides this module and how it heavily use class variables I think that something is bad with design of this module (too many @@, no instance variable, instance methods doesn't use instance data (because noone is present)). When I think about this module I see this solution as the best one: Use singleton class which manages background process and have hash of running, done and mutex as instance variables > + # > + # currently running requests in background (current states) > + @@running = Hash.new > + # finished requests (actual results) > + @@done = Hash.new > + # a mutex which guards access to the shared class variables above > + @@mutex = Mutex.new > + > + # define a new background process > + # id is unique ID > + def add_process(id) > + @@mutex.synchronize do > + @@running[id] = BackgroundStatus.new unless @@running.has_key?(id) > + end > + end > + > + # is the process running? > + def process_running?(id) > + @@mutex.synchronize do > + return @@running.has_key? id This doesn't work. You cannot return from from block ( I already study this problem when catch problem in validates_each ). Because it return data in block which return statement could be ignored ( because synchronize do lock; yield,unlock so even if this work it return value from unlock - if the most case it is expected value but can be nil see below). just try it in irb: m.synchronize { return 'a' } LocalJumpError: unexpected return from (irb):5 from (irb):5:in `synchronize' from (irb):5 from :0 so use instead variable ret = false @@mutex.synchronize do ret = @@running.has_key? id end return ret # I personally prefer explicit return even if it is not needed in ruby This lead me to question - is this code tested? It raise exception each time, so it is strange for that it is not discovered (I found my problem with validation in tests). > + end > + end > + > + # is the process finished? > + def process_finished?(id) > + @@mutex.synchronize do > + return @@done.has_key? id ^^^ same as above > + end > + end > + > + # remove the progress status and remember the real final value > + def finish_process(id, value) > + @@mutex.synchronize do > + @@running.delete(id) > + @@done[id] = value > + end > + end > + > + # get the current progress > + # returns a copy, use update_progress() for updating the progress > + def get_progress(id) > + @@mutex.synchronize do > + ret = @@running[id] > + ret.nil? ? nil : ret.dup ^^^ This return excepted value if you not make critical mistake ( because it return nil, if you unlock mutex on which you invoke synchronize method. So I recommend checking return value of synchronize to detect locking problem (but you must change inside block to not return nil in any case). so something like : ret = nil mret = @@mutex.synchronize do ret = @@running[id] ret = ret.dup unless ret.nil? "OK" done raise "Someone unlock mutex during executing synchronize" unless mret return ret working in threads and locking need special care and study how it really work, otherwise some really ugly problem can raise. > + end > + end > + > + # get the final value, the value is removed from the internal structure > + def get_value(id) > + @@mutex.synchronize do > + return @@done.delete id ^^^ same as above > + end > + end > + > + # update the progress > + def update_progress(id, &block) > + @@mutex.synchronize do > + bs = @@running[id] > + > + yield bs if bs && block_given? > + end > + end > + > + # is background processing possible? > + # if cache_classes is disabled it is not possible > + # because all classes are reloaded between requests > + # and the static attributes keeping the progress are lost > + def background_enabled? > + return Rails.configuration.cache_classes > + end > + > +end > + > When I check this commit I also found some not bad practice in BackgroundStatus: attr_reader :status def status=(status) if @status != status @status = status trigger_callback end end I don't like it, because it is confusing use same name for parameter and for attribute. There is confusing if status is status reader or argument (of course if you know how in deep ruby work ,then parameter as local variable overwrite instance method, but I find it confusing. and little hints: def trigger_callback if @callback @callback.call end end you could write it as def trigger_callback @callback.try(:call) end (or if you don't want use rails object extension you can still write it as oneliner) you don't provide to_json method which could be problem if we switch frontend to json. There you can benefit from serialization module which I write for BaseModel. So instead of whole to_xml method you can get same result (and as bonus also to_json and from_xml and from_json method) if you use this code: (attr_serialized is needed to avoid callback instance serialization) include BaseModel::Serialization attr_serialized :status, :progress, :subprogress Note - all code is code from head, so no guaranty that it works out of box, but it should work with little fixes. Josef -- Josef Reidinger YaST team maintainer of perl-Bootloader, YaST2-Repair, webyast (language,time,basesystem,ntp) -- To unsubscribe, e-mail: yast-devel+unsubscribe(a)opensuse.org For additional commands, e-mail: yast-devel+help(a)opensuse.org

1 2

[yast-devel] Webyast: ancestor for models in rest-service merged to master
by Josef Reidinger 14 Dec '09

14 Dec '09

Hi, I implement ancestor which have some usable utilities for rest-service models. Reason to have it is in bug http://bugzilla.novell.com/show_bug.cgi?id=543423 . I call it BaseModel and it is inspired by ActiveModel from rails3.0. It is general usability and I think that all models which is not based on database or REST service could benefit from it. ( just inform me, I can help you to use it in your project). Main features is: * callbacks (before_save etc.) * validations ( validates_format_of etc.) * serialization ( to_xml and to_json and also restore it) * mass assignment (load instance from hash, also whitelisting and blacklisting attributes for mass assignment). For details see documentation: http://yast.suse.de:3334/view/webyast/job/webyast-service-head- i586/javadoc/webservice/index.html classes in BaseModel namespace (sorry internal only, as we still doesn't have public place where we synchronize documentation ). Or you can generate it in git by rake doc:app in webservice directory. For example usage see time plugin, which use validations, after_save hook, mass assignment and also serialization ( It is also example of xml limitation as for hash is key used as tags and tags is quite restricted, so time module must transform hash before serialization ). Welcome your comments, suggestions or ideas what could be improved or added ( especially if more then one plugin could benefit from it). Josef -- Josef Reidinger YaST team maintainer of perl-Bootloader, YaST2-Repair, webyast (language,time,basesystem,ntp) -- To unsubscribe, e-mail: yast-devel+unsubscribe(a)opensuse.org For additional commands, e-mail: yast-devel+help(a)opensuse.org

1 0

[yast-devel] WebYaST feedback
by Martin Vidner 14 Dec '09

14 Dec '09

Hi, Stano has already entered the individual problems to bugzilla, but I thought it would be useful to summarize them - Customization is lacking Style sheets are kind of messy. It is hard to replace "My Appliance" with "ACME Foobarizer". Help cannot be customized. Oh wait, there is no help. - Module offer is enough For Foobarizer, his first boot would contain License Root password (maybe Network) Registration He does not expect a need to write a new module for the first boot. But if he needs to, he would write it in their application's framework, not in WebYaST. -- Martin Vidner, YaST developer http://en.opensuse.org/User:Mvidner Kuracke oddeleni v restauraci je jako fekalni oddeleni v bazenu

2 1

[yast-devel] [ANNOUNCE] ZYpp source repositories on gitorious.org/opensuse
by Duncan Mac-Vicar Prett 11 Dec '09

11 Dec '09

The ZYpp repositories previously published at git.opensuse.org repositories are now available on http://gitorious.org/opensuse http://gitorious.org/opensuse/sat-solver http://gitorious.org/opensuse/libzypp http://gitorious.org/opensuse/zypper http://gitorious.org/opensuse/zypp-utils (previously zypp/tools) http://gitorious.org/opensuse/libzypp-bindings http://gitorious.org/opensuse/libzypp-testsuite http://gitorious.org/opensuse/zypp-dialogsolver http://gitorious.org/opensuse/cmpi-zypp http://gitorious.org/opensuse/zypp-zmd-backend (previously zypp/backend) There is an initial opensuse-zypp team defined with access to those repositories. To switch you only need to edit your .git/config origin remote If before it was git@git.opensuse.org:projects/zypp/zypper.git Now it would be: git@gitorious.org:opensuse/zypper.git The repositories in git.opensuse.org are still accessible, but you can't push to them anymore. They will stay there until no repositories are left on that machine. Gitorious provides feeds for each repository pushes, and per each branch commits. To make it easier to follow the overall project, I have created this feed which merges various feeds and make the titles more similar to the old mails: http://pipes.yahoo.com/pipes/pipe.info?_id=36dc1dc1c0380cca92df9ddaa5b6e26d In the next days we will look for a solution to mail those feed updates to the commits mailing list. -- Duncan Mac-Vicar P. - Engineering Manager, YaST SUSE LINUX Products GmbH, GF: Markus Rex, HRB 16746 (AG Nuernberg) -- To unsubscribe, e-mail: yast-devel+unsubscribe(a)opensuse.org For additional commands, e-mail: yast-devel+help(a)opensuse.org

1 0

[yast-devel] Nokogiri branch
by Josef Reidinger 11 Dec '09

11 Dec '09

Hi, I create use_nokogiri branch for webclient. My quick test doesn't show any problem. Could you try it and report me if any problem occur in your modules [1]? Nokogiri advantage is better performance without almost any cost. thanks Josef [1] howto run nokogiri branch: zypper in rubygem-nokogiri git checkout -t origin/use-activeresource #in webclient directory start.sh #start webclient -- Josef Reidinger YaST team maintainer of perl-Bootloader, YaST2-Repair, webyast (language,time,basesystem,ntp) -- To unsubscribe, e-mail: yast-devel+unsubscribe(a)opensuse.org For additional commands, e-mail: yast-devel+help(a)opensuse.org

2 2

[yast-devel] questions concerning yast2-testsuite
by Arvin Schnell 11 Dec '09

11 Dec '09

Hi, I have some problems with the yast2 testsuite. When I run tests in storage y2base cannot find the libstorage bindings since they are not installed in the system. So I need to modify the Makefile.am get unfortunately gets overwritten by the testsuite. Is there any mechanism to handle specific changes to the files from the testsuite skeleton? Also I would like to change testsuite.exp to run all tests even after a failure. Objections? -foreach file $filenames { - if {[testsuite-run $file] < 0} { - return - } -} +foreach file $filenames { testsuite-run $file } caio Arvin -- Arvin Schnell, <aschnell(a)suse.de> Senior Software Engineer, Research & Development SUSE LINUX Products GmbH, GF: Markus Rex, HRB 16746 (AG Nürnberg) -- To unsubscribe, e-mail: yast-devel+unsubscribe(a)opensuse.org For additional commands, e-mail: yast-devel+help(a)opensuse.org

3 5

[yast-devel] wiki - updated Rails debugging page
by Ladislav Slezak 11 Dec '09

11 Dec '09

Hi all, I have updated the Rails debugging page (http://en.opensuse.org/YaST/Web/Development/Tips) Added sections: * starting Rails app directly in the ruby debugger * debugging a Ruby process using GDB debugger - if the app does not respond at all you can still get the backtrace or introspect the objects (actually it's just a link to http://eigenclass.org/hiki.rb?ruby+live+process+introspection ) -- Best Regards Ladislav Slezák Yast Developer ------------------------------------------------------------------------ SUSE LINUX, s.r.o. e-mail: lslezak(a)suse.cz Lihovarská 1060/12 tel: +420 284 028 960 190 00 Prague 9 fax: +420 284 028 951 Czech Republic http://www.suse.cz/ -- To unsubscribe, e-mail: yast-devel+unsubscribe(a)opensuse.org For additional commands, e-mail: yast-devel+help(a)opensuse.org

1 0

[yast-devel] WebYaST status call 11-Dec-2009
by Jiří Suchomel 11 Dec '09

11 Dec '09

participants: jsuchome, kkaempf, mvidner, jsrain, schubi, lslezak, jdsn, dmacvicar mvinder: plan to learn new things, encourages others lslezak: testing and preparing patches patch, seen DBUS problems kkaempf: working on announcements, release notes, until noon schubi: fixing bugs in branch, looking for status page improvement dmacvicar: research on autocompletion, validation Next call: 4th January 2010 -- Jiri Suchomel SUSE LINUX, s.r.o. e-mail: jsuchome(a)suse.cz Lihovarská 1060/12 tel: +420 284 028 960 190 00 Praha 9, Czech Republic http://www.suse.cz -- To unsubscribe, e-mail: yast-devel+unsubscribe(a)opensuse.org For additional commands, e-mail: yast-devel+help(a)opensuse.org

1 0

[yast-devel] Webyast Serialize hash to xml
by Josef Reidinger 11 Dec '09

11 Dec '09

Hi, when I work on ancestor for rest-service I found interesting problem. How to serialize hash with general keys. Problem is that key could contain prohibited characters and there is no way how to escape it in xml tag (in json it works without problems). I see few solution and want discuss which one is preferred. 1) prohibit general keys and allow only developer defined string: It is how it is done now, as we manually create xml and hash is encoded to different tags ( so eg replace general key => value with [ :key => key, :value => value] ) (it is solved e.g. in time where is map with id_of_timezone => name of timezone) advantage: - easy to do - same as we did it now - easy to use for customer which want use directly rest-service disadvantage: - key must be checked otherwise it could be security problem - more work in rest-service and webclient as hash must be transformer before and after serialization 2) use inlined yaml - serialize hash to yaml and inline to xml. It escaped any problematic patterns advantage: - work out of box - easy to use - no additional transformation disadvantage: - problem for customer which want to use directly rest-service as he must define hash as yaml - on webclient hash must be saved as yaml (easy call hash.to_yaml) which is unintuitive 3) hack webclient Hash.from_xml. I can modify hash.from_xml behavior so if he get hash type it expects array with tags key and value, so result looks same as in 1) and hack it also on backend so it is automatic recognize and just works. advantage - - work out of box - developer doesn't know about it, just works disadvantage - - if customer use nonmodified rails, then it doesn't read correctly as hash (but still can work with it) - need modifications in webclient and rest-service xml serialization and deserialization 4) Use only json - Provide only json in rest-service advantage: - json has better performance then xml, as it is easier to parse and create disadvantage: - no XML, so customer could complain I welcome any comments about solutions or propose of new solution. Josef -- Josef Reidinger YaST team maintainer of perl-Bootloader, YaST2-Repair, webyast (language,time,basesystem,ntp) -- To unsubscribe, e-mail: yast-devel+unsubscribe(a)opensuse.org For additional commands, e-mail: yast-devel+help(a)opensuse.org

1 0

[yast-devel] a few notes on autocompletion
by Duncan Mac-Vicar Prett 11 Dec '09

11 Dec '09

After playing some days with forms and the users module in order to achieve the following goals: - autocompletion of form data - examples for fields - browser side validation I want to share some results of those experiments and how they affect our design The first limitation I found is mostly likely present in various models. Users.find(:all) returns less data than Users.find(id), only the username and something else. If you want to autocomplete/suggest all known login shells ( something like Users.find(:all).map { |x| x.login_shell } ) you need a way on the client side to get all login shells. I ended modifying the /users service to - accept ?attributes[]=attr1&attributes[]=attr2... so that the caller could chose which attributes to insert - be sure that it was able to return json Once you have that, doing autocompletion is trivial, but very repeat-yourself prone. Biggest problem is that you can't call from the browser the rest- service unless: - you set the url of the resource in the view - you use JSON-P, using $.getJSON and setting a jsonp callback ( http://en.wikipedia.org/wiki/JSON#JSONP ) Or you can create an action in the controller which calls the service and returns the result. I can't get the jsonp call to actually process the data. Then you need to set the right autocomplete data in the javascript document.ready callback inlined in the view (at least until we solve the problem I mentioned in opensuse-ruby mailing list ), by using jquery.autocomplete calls on the right form fields, giving the autocomplete data you got from the service or the controller action We can avoid duplication of Javascript by emulating what rails implements with protoype, so that you specify the autocompletions in the model class. For that we would need to swtich from find_proxy to a real User class using the class Martin Kudlvasr and Martin Vidner where working on. Once you have a model we could use something like: http://github.com/grosser/simple_auto_complete combined with the jquery.autocomplete plugin to achieve a more or less standard to apply validation to our forms. Any ideas? -- Duncan Mac-Vicar P. - Engineering Manager, YaST SUSE LINUX Products GmbH, GF: Markus Rex, HRB 16746 (AG Nuernberg) -- To unsubscribe, e-mail: yast-devel+unsubscribe(a)opensuse.org For additional commands, e-mail: yast-devel+help(a)opensuse.org

2 1

[yast-devel] WebYaST GMC (gold master candidate)
by Martin Vidner 10 Dec '09

10 Dec '09

0. Introduction =============== WebYaST is a web-based remote console plugging into a stateless interface exposed by an SLE-based appliance and providing a set of configuration modules appropriate to enable minimal user control of the appliance "black box" environment. 1. Installation =============== WebYaST is not "installable" per se. It is a component available for building of appliances. To build appliances using webYaST, set up the repository in your SUSE Studio instance. See also these documentation packages for vendors: webyast-vendor_en.rpm webyast-vendor_en-pdf.rpm REPOSITORY: zypper ar http://www.suse.de/~mvidner/webyast-gmc webyast-gmc zypper ar http://www.suse.de/~mvidner/webyast-lighty webyast-lighty (webyast-lighty contains fixes for SLES and SDK which will appear as maintenance updates for these base products) APPLIANCE DOWNLOAD: http://www.suse.de/~mvidner/webyast-gmc/vm/Signed_WebYaST_on_SLES11.i686-0.… http://www.suse.de/~mvidner/webyast-gmc/vm/Signed_WebYaST_on_SLES11.i686-0.… ( gpg --verify Signed_WebYaST_on_SLES11.i686-0.26.0.vmx.tar.gz.asc ) sha1sum: a79fb9c417eee226d9199403d0cbf02545a5cc64 internal: ftp://cml.suse.cz/testing/webyast/Signed_WebYaST_on_SLES11.i686-0.26.0.vmx.tar.gz MODULE SPECIFIC NEWS: http://www.suse.de/~mvidner/webyast-gmc/ChangeLog 2. Usage ======== Run our sample appliance. The WebYaST console is running at port 54984 in the resulting appliance. Point the browser to https://<appliance-hostname>:54984/ (**) Log in as root, with password linux yast2-webservice-basesystem is included, go through the base setup (language, eulas, admin, network, time, mail, registration) See also these documentation packages for users: webyast-user_en.rpm webyast-user_en-pdf.rpm 3. Known Issues =============== No P1, 4 P2s: 559145,"Major","mvidner(a)novell.com","schubi(a)novell.com","NEW","Changed Domain name not reflected when sending mail" 560701,"Major","mvidner(a)novell.com","jsuchome(a)novell.com","NEW","[build 24.1] webYaST appears to crash after installing webclient patch" 560292,"Normal,"dmacvicar(a)novell.com","schubi(a)novell.com","NEW","Error while updating yast2-webclient package" 561642,"Normal","lslezak(a)novell.com","thomas(a)novell.com","ASSIGNED","WebYAST RC4 webservice mail_settings fuzzing causes dozens of sh processes to hang around" See this query for a complete bug list: https://bugzilla.novell.com/buglist.cgi?query_format=advanced&product=openS… 4. Feedback =========== We are eager to get your feedback. Please, report the bugs, post general comments and questions to the yast-devel(a)opensuse.org mailing list. Other alternatives: - irc://irc.opensuse.org/yast - https://bugzilla.novell.com/enter_bug.cgi?component=WebYaST&product=openSUS… (please disregard the oS11.2 version and instead remember to mention the appliance version.) See http://en.opensuse.org/Bugs/WebYaST - mailto:yast-devel@opensuse.org -- Martin Vidner, WebYaST release engineer http://www.youtube.com/watch?v=PDwH9UlRIjc 3... 2... 1...

5 8

[yast-devel] WebYaST status call 10-Dec-2009
by Jiří Suchomel 10 Dec '09

10 Dec '09

participants: jdsn, jsuchome, jsrain, mvidner, lslezak, schubi, dmacvicar all: testing mvidner: GMC released in time (before the party) lslezak: fixed bnc #561642, not for GM schubi: checking permissions, work in branch -- Jiri Suchomel SUSE LINUX, s.r.o. e-mail: jsuchome(a)suse.cz Lihovarská 1060/12 tel: +420 284 028 960 190 00 Praha 9, Czech Republic http://www.suse.cz -- To unsubscribe, e-mail: yast-devel+unsubscribe(a)opensuse.org For additional commands, e-mail: yast-devel+help(a)opensuse.org

1 0

[yast-devel] WebYaST status call 09-Dec-2009
by Jiří Suchomel 09 Dec '09

09 Dec '09

participants: jsuchome, jdsn, jsrain, mvidner, jreidinger, kkaempf, dmacvicar all: testing jreidinger: dbus models ancestor, work in branch mvidner: package kit bug fix, build problems passed, working on final release dmacvicar: moving to gitorious: see yast-internal after release: check bugzilla bugs, set priorities, 11.2 status not clear yet -- Jiri Suchomel SUSE LINUX, s.r.o. e-mail: jsuchome(a)suse.cz Lihovarská 1060/12 tel: +420 284 028 960 190 00 Praha 9, Czech Republic http://www.suse.cz -- To unsubscribe, e-mail: yast-devel+unsubscribe(a)opensuse.org For additional commands, e-mail: yast-devel+help(a)opensuse.org

1 0

[yast-devel] Profiling rest-service
by Josef Reidinger 08 Dec '09

08 Dec '09

Hi, I profile also rest-service. Its result is quite good. I choose time plugin index action. It shows that it spend the most of time in building XML. DBus layer takes just few percents (around 3%). So bottleneck of rest-service (beside YaPI) is creating XML, but I don't know better alternative. Josef -- Josef Reidinger YaST team maintainer of perl-Bootloader, YaST2-Repair, webyast (language,time,basesystem,ntp) -- To unsubscribe, e-mail: yast-devel+unsubscribe(a)opensuse.org For additional commands, e-mail: yast-devel+help(a)opensuse.org

2 2

[yast-devel] WebYaST status call 07-Dec-2009
by Jiří Suchomel 08 Dec '09

08 Dec '09

WebYaST status call 07-Dec-2009 participants: jsuchome, jdsn, mvidner, schubi, jsrain, jreidinger, mschmidkunz, dmacvicar, bbrunner, visnov jsuchome: testing lighty bug fix jdsn: testing, building appliance mvidner: Friday fixes are ready, but not sufficient (sysvinit). Should we ship? Yes, after this bug is solved. schubi: testing, looking for other possible solutions for lighttpd jreidinger: client profiling: Rails XML parser slow - some promising results mschmidkunz: testing with RC4, talk with Garret about jquery dmacvicar: mail about UI review, see yast-internal -- Jiri Suchomel SUSE LINUX, s.r.o. e-mail: jsuchome(a)suse.cz Lihovarská 1060/12 tel: +420 284 028 960 190 00 Praha 9, Czech Republic http://www.suse.cz -- To unsubscribe, e-mail: yast-devel+unsubscribe(a)opensuse.org For additional commands, e-mail: yast-devel+help(a)opensuse.org

2 1

[yast-devel] Re: WebYaST on 11.2
by Martin Vidner 08 Dec '09

08 Dec '09

On Mon, Dec 07, 2009 at 10:50:54PM +0000, Tim Schuermann wrote: > Hi Martin! > > Which package I have to install? > > (rubygem-rails-2_3 is already installed at my openSUSE 11.2-installation.) > > Regards, > > Tim Hi Tim, I think there is some other rails dependency missing, but due to the bug below it is not so easy to know which. You will know once you resolve the sysvinit dependency. You need sysvinit from http://download.opensuse.org/repositories/Base://System/openSUSE_Factory and lighttpd, cannot find it now :-( https://bugzilla.novell.com/show_bug.cgi?id=559534#c69 > > See http://bugzilla.novell.com/show_bug.cgi?id=561392 for > > the > > details. -- Martin Vidner, YaST developer http://en.opensuse.org/User:Mvidner Kuracke oddeleni v restauraci je jako fekalni oddeleni v bazenu

1 0

[yast-devel] WebYaST on 11.2
by Martin Vidner 07 Dec '09

07 Dec '09

Hi, on IRC we looked at a problem where WebYaST on 11.2 gives only a minimal 500 page. I haven't resolved all issues yet, but one problem is that ignoring a lighttpd dependency ignored also entire Rails :-) See http://bugzilla.novell.com/show_bug.cgi?id=561392 for the details. -- Martin Vidner, YaST developer http://en.opensuse.org/User:Mvidner Kuracke oddeleni v restauraci je jako fekalni oddeleni v bazenu

1 0

[yast-devel] Profiling web-client
by Josef Reidinger 07 Dec '09

07 Dec '09

Hi, I profile web-client to find where is bottleneck. I run tests on network plugin (index action) as it takes the most serious amount of time. To profile I use ruby-prof in specified part of code. 1) almost all time takes internal parsing in REXML (more then 80%) (waiting on rest-service is not measured). 2) permissions recognition take around 7% of time. 3) resource_for_interface takes around 20% of time. comments: add 2) I create bug - https://bugzilla.novell.com/show_bug.cgi?id=561265 . It should be easily cached, as permission for target host is not changed often. So on login it could be stored in local db information about permissions. add 3) I create bug -https://bugzilla.novell.com/show_bug.cgi?id=561283 . It can be cached as well, as routes for interface is not changed. So at least during login it should be stored and then used. add 1) I read some articles and there is much better alternatives then default REXML. The easiest usage and good support in rails has nokogiri. I try it and it is more then 3x faster ( 960k cycles REXML versus 290k cycles for nokogiri). I try it also visually and it looks little faster (flickr says that its speedup is 40%). You can also try it. Just install rubygem-nokogiri, then add this line to your enviroment.rb: ActiveSupport::XmlMini.backend = 'Nokogiri' And it works (I am not sure if this package is in SLE11. How can I find it?). Josef -- Josef Reidinger YaST team maintainer of perl-Bootloader, YaST2-Repair, webyast (language,time,basesystem,ntp) -- To unsubscribe, e-mail: yast-devel+unsubscribe(a)opensuse.org For additional commands, e-mail: yast-devel+help(a)opensuse.org

2 2

[yast-devel] WebYaST status call 03-Dec-2009
by Jiri Srain 03 Dec '09

03 Dec '09

participants: jsrain, mvidner, schubi, lslezak, kkaempf, dmacvicar, visnov, mkudlvasr jsrain - confirmed that lighttpd fix did not solve the problem mvidner - working on lighttpd issue and image building shubi - lighttpd bug lslezak - testing, looking into bugs - looking for performance bottlenecks, identified calls to optimize kkaempf - lighttpd bug mkudlvasr - testing RC7, researching caching in Rails mkudlvasr + lslezak TODO: look what can get broken on package update -- To unsubscribe, e-mail: yast-devel+unsubscribe(a)opensuse.org For additional commands, e-mail: yast-devel+help(a)opensuse.org

1 0

[yast-devel] Is it allowed to modify one same list via foreach?
by Johannes Meixner 02 Dec '09

02 Dec '09

Hello, is such code allowed: -------------------------------------------------------------------- list < string > words = [ "Jane", "John" ]; integer index = -1; foreach( string word, words, { index = index + 1; words[index] = "Hello " + word; } ); -------------------------------------------------------------------- Is in any case the result words == [ "Hello Jane", "Hello John" ] ? If the above is allowed, what about even more evil looking code: -------------------------------------------------------------------- list < string > words = [ "Jane", "John", "Joe" ]; integer index = -1; foreach( string word, words, { index = index + 1; if( "John" == word ) { remove( words, index ); } } ); -------------------------------------------------------------------- Is in any case the result words == [ "Jane", "Joe" ] ? If it is not allowed, how to copy a list? Does -------------------------------------------------------------------- list < string > words = [ "Jane", "John", "Joe" ]; list < string > copy = words; -------------------------------------------------------------------- result two totally separated lists words and copy or does copy contain only a reference which points to words so that -------------------------------------------------------------------- remove( copy, 1 ); -------------------------------------------------------------------- would actually result copy == words == [ "Jane", "Joe" ] ? Kind Regards Johannes Meixner -- SUSE LINUX Products GmbH, Maxfeldstrasse 5, 90409 Nuernberg, Germany AG Nuernberg, HRB 16746, GF: Markus Rex -- To unsubscribe, e-mail: yast-devel+unsubscribe(a)opensuse.org For additional commands, e-mail: yast-devel+help(a)opensuse.org

3 3

[yast-devel] WebYaST status call 02-Dec-2009
by Jiri Srain 02 Dec '09

02 Dec '09

participants: jsrain, jdsn, jsuchome, schubi, mvidner, kkaempf, mschidkunz, dmacvicar everybody - testing - last image did not solve the key issue jsuchome - found solution of mail problem with Petr Varkoly, suggesting wait for maintenance update - observed crash after upgrade of lighttpd schubi - will have a look at the lighttpd problem mvidner - created initial release notes - will also have a look at the lighttps problem -- To unsubscribe, e-mail: yast-devel+unsubscribe(a)opensuse.org For additional commands, e-mail: yast-devel+help(a)opensuse.org

1 0

[yast-devel] yast2-gtk sw_single thread
by Ricardo Cruz 01 Dec '09

01 Dec '09

Hi there, There is an ongoing discussion over the future of yast2-gtk's software manager UI at: http://groups.google.com/group/yast2-gtk Feel free to subscribe and throw your 2 cents. Cheers, Ricardo ---------------------------------------------------------------- This message was sent using IMP, the Internet Messaging Program. -- To unsubscribe, e-mail: yast-devel+unsubscribe(a)opensuse.org For additional commands, e-mail: yast-devel+help(a)opensuse.org

1 0

[yast-devel] WebYaST status call 01-Dec-2009
by Jiri Srain 01 Dec '09

01 Dec '09

participants: jsrain, visnov, mvidner, lslezak, kkaempf, jdsn, dmacvicar, schubi, jsuchome, mschmidkunz mvidner - built RC6, still reports "No Updates" - keys are not imported, UI does not detect this error - release notes are missing lslezak - testing, debugging some bugs in classic YaST which may affect WebYaST and lead into infinite loop (actually not affecting) kkaempf - busy with effort estimates - we have to fix patch installation (the only blocker) jdsn - various tests, found missing GPG keys schubi - testing - will look into patch stuff jsuchome - testing & fixing esp. mail settings, still no idea what's wrong with mail everybody - testing latest builds, working on bugfixes - we need to verify that documenting how to add keys is sufficient -- To unsubscribe, e-mail: yast-devel+unsubscribe(a)opensuse.org For additional commands, e-mail: yast-devel+help(a)opensuse.org

2 1