On Wed, Aug 06, 2008 at 04:57:13PM +0200, Michael Andres wrote:

On Wed, Aug 06, Martin Vidner wrote:

...

ALog::Item("/etc/ntp.conf: added 'server ntp.example.org'") ALog::Item("enabled /etd/init.d/ntp") ALog::Item("started /etd/init.d/ntp")

ALog::CommitPopup() will lead to ALog::Note("set up ntp from local server")

Isn't this what the loglevel 6 (E_USR) was intended for? An absract of changes done to the system which is also loged into a separate file.

Yes. The basic requirement is simple: better logging. Naturally the design is simple too: a function for the better logging. (I put it into a namespace to make room for its brother, a commit message from the admin.) E_USR and analogically y2user(fmt, ...) seems to imply the constraints shared by the other logging functions: the same log file, the same format... I forgot to mention that I assume it will use a different log file (and additionally use y2log too probably) so that people do not have to call grep to make sense of it. (E_USR is in http://svn.opensuse.org/svn/zypp/trunk/libzypp/zypp/base/Logger.h ) -- Martin Vidner, YaST developer http://en.opensuse.org/User:Mvidner Kuracke oddeleni v restauraci je jako fekalni oddeleni v bazenu -- To unsubscribe, e-mail: yast-devel+unsubscribe@opensuse.org For additional commands, e-mail: yast-devel+help@opensuse.org

On Wed, Aug 06, 2008 at 05:53:58PM +0200, Arvin Schnell wrote:

On Wed, Aug 06, 2008 at 05:41:49PM +0200, Martin Vidner wrote:

...

On Wed, Aug 06, 2008 at 04:57:13PM +0200, Michael Andres wrote:

...

Isn't this what the loglevel 6 (E_USR) was intended for? An absract of changes done to the system which is also loged into a separate file.

Yes. The basic requirement is simple: better logging. Naturally the design is simple too: a function for the better logging. (I put it into a namespace to make room for its brother, a commit message from the admin.) E_USR and analogically y2user(fmt, ...) seems to imply the constraints shared by the other logging functions: the same log file, the same format...

Log file and format should be possible to change.

I'm not sure what you mean. During the design or at runtime?

The problem with ALog::Item is that it doesn't understand %1 syntax. Always using sformat is clumsy.

Right. It seems I will have to improve core to make it possible.

Also ALog::Item is not accessible from C++ (in case anybody needs that). Access from libstorage seems idealistic since libstorage uses limal/blocxx stuff for logging.

With some work, both should be possible. What other C++ users aside from libstorage would we have? Remember ALog as a YCP module is just a prototype. -- Martin Vidner, YaST developer http://en.opensuse.org/User:Mvidner Kuracke oddeleni v restauraci je jako fekalni oddeleni v bazenu -- To unsubscribe, e-mail: yast-devel+unsubscribe@opensuse.org For additional commands, e-mail: yast-devel+help@opensuse.org

On Thu, Aug 07, 2008 at 03:20:53PM +0200, Martin Vidner wrote:

On Wed, Aug 06, 2008 at 05:53:58PM +0200, Arvin Schnell wrote:

...

On Wed, Aug 06, 2008 at 05:41:49PM +0200, Martin Vidner wrote:

...

On Wed, Aug 06, 2008 at 04:57:13PM +0200, Michael Andres wrote:

...

Isn't this what the loglevel 6 (E_USR) was intended for? An absract of changes done to the system which is also loged into a separate file.

Yes. The basic requirement is simple: better logging. Naturally the design is simple too: a function for the better logging. (I put it into a namespace to make room for its brother, a commit message from the admin.) E_USR and analogically y2user(fmt, ...) seems to imply the constraints shared by the other logging functions: the same log file, the same format...

Log file and format should be possible to change.

I'm not sure what you mean. During the design or at runtime?

I mean it should be easy to implement a new log macro y2user in C++ that logs into a different file with a different format. Accessing that from YCP would also be easy and you get the "%1" things for free. ciao Arvin -- To unsubscribe, e-mail: yast-devel+unsubscribe@opensuse.org For additional commands, e-mail: yast-devel+help@opensuse.org

* Arvin Schnell [Aug 09. 2008 13:58]:

Callbacks would be another possibility. But I suppose logging is also required when libstorage is used by other programs so callbacks would only move the problem to those programs.

Implementing (optional) logging callbacks is a smaller 'problem' than being forced to use a specific, fixed logging implementation. Looking at using libstorage in other programs, I'd prefer logging callbacks. Klaus -- To unsubscribe, e-mail: yast-devel+unsubscribe@opensuse.org For additional commands, e-mail: yast-devel+help@opensuse.org

On Wed, Aug 06, 2008 at 10:33:01PM +0200, Michal Svec wrote:

Can't we get this information from SCR, without calling explicitely some functions (which is going to be error prone)?

For instance the example above could be generated from SCR (layer), one would (more or less :) only need .path -> file/function translation to create such a log.

SCR could do it if that was a reuqirement from early on, but I am afraid that it is not quite suited for this use at this point. In other words, it would be rather "much more" than "less" I think. Anyway, we can see how that would look like from watching what goes through SCR: Y2DEBUG=1 (or [Debug] scr=true in log.conf), then egrep -A3 'executeSubagentCommand: (Write|Execute)' y2log For example attached is one such log, which I would like to see described as Note "changed ntp server from ntp.suse.cz to random one from pool.ntp.org" Item "/etc/ntp/conf: server list changed to 0.pool.ntp.org 1.pool.ntp.org 2.pool.ntp.org" Item "/etc/sysconfig/ntp: NTPD_RUN_CHROOTED changed to yes" (hmm, that's a gratuitous change) Item "restarted /etc/init.d/ntp" I encourage people to try this with their own modules and perhaps post the results so that we get a better idea if pursuing this path is worthwhile. -- Martin Vidner, YaST developer http://en.opensuse.org/User:Mvidner Kuracke oddeleni v restauraci je jako fekalni oddeleni v bazenu

Michal Svec wrote:

On Thu, 7 Aug 2008, Martin Vidner wrote:

...

On Wed, Aug 06, 2008 at 10:33:01PM +0200, Michal Svec wrote:

...

Can't we get this information from SCR, without calling explicitely some functions (which is going to be error prone)?

For instance the example above could be generated from SCR (layer), one would (more or less :) only need .path -> file/function translation to create such a log.

SCR could do it if that was a reuqirement from early on, but I am afraid that it is not quite suited for this use at this point. In other words, it would be rather "much more" than "less" I think.

Sure, this was just an idea.

...

Anyway, we can see how that would look like from watching what goes through SCR: Y2DEBUG=1 (or [Debug] scr=true in log.conf), then egrep -A3 'executeSubagentCommand: (Write|Execute)' y2log

It actually looks somehow similar (up to slight syntax differences) to what you suggested, see the attached output from one run of the yast2 security module.

See also this thread: # From: "Alexey Eremenko" # Date: Tue, 29 Jul 2008 14:57:24 +0000 # [opensuse] Yast as a Linux learning tool http://lists.opensuse.org/opensuse/2008-07/msg02571.html L.