[opensuse-packaging] Removing SLE 11 from devel:languages:python
I propose we remove the SLE 11 repository from devel:languages:python. Most spec files require special workaround to build at all with this target and it requires backporting a bunch of system-critical packages that no other release requires, and even then many packages fail to build. And the scheduler punishes the repo for having failures. The amount of effort needed to keep basic, key dependencies working is growing constantly. And these workarounds means that it will be essentially infeasible to support it in the upcoming single spec file system, which means most key packages will no longer be able to support it anyway. Anyone who really needs specific packages can always link to them and build them in their home repo. So I think we should go ahead and remove it. Once it is removed, packages that are updated can have the workarounds deleted -- To unsubscribe, e-mail: opensuse-packaging+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-packaging+owner@opensuse.org
On Tue, 2016-11-22 at 10:52 -0500, Todd Rme wrote:
I propose we remove the SLE 11 repository from devel:languages:python.
For what is worth, I agree. -- SUSE Linux GmbH, GF: Felix Imendörffer, Jane Smithard, Graham Norton, HRB 21284 (AG Nürnberg) Maxfeldstraße 5, 90409 Nürnberg, Germany -- To unsubscribe, e-mail: opensuse-packaging+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-packaging+owner@opensuse.org
On 22 November 2016 at 16:59, Alberto Planas Dominguez
On Tue, 2016-11-22 at 10:52 -0500, Todd Rme wrote:
I propose we remove the SLE 11 repository from devel:languages:python.
For what is worth, I agree.
For what it's worth, I agree also -- To unsubscribe, e-mail: opensuse-packaging+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-packaging+owner@opensuse.org
On 11/22/2016 10:52 AM, Todd Rme wrote:
I propose we remove the SLE 11 repository from devel:languages:python. Most spec files require special workaround to build at all with this target and it requires backporting a bunch of system-critical packages that no other release requires, and even then many packages fail to build. And the scheduler punishes the repo for having failures.
The amount of effort needed to keep basic, key dependencies working is growing constantly. And these workarounds means that it will be essentially infeasible to support it in the upcoming single spec file system, which means most key packages will no longer be able to support it anyway.
Anyone who really needs specific packages can always link to them and build them in their home repo.
So I think we should go ahead and remove it. Once it is removed, packages that are updated can have the workarounds deleted
I agree that it is a pain. I also agree that going forward things will become ever more painful. However the approach of "link to the project" where needed is not really a good solution either. If we drop SLE_11 builds and then remove the special SLE_11 handling in the spec file the person that has linked to the package MUST carry a local diff in addition to the link to re-add the SLE_11 special handling. Chances are that the local diff will be broken every time the package in d:l:p gets updated. Consequently people will probably make copies meaning a possibly significant increase of builds in OBS. So if we turn of SLE_11 in d:l:p we should agree to keep the SLE_11 special code in the spec files until EOL of SLE_11, sometime in 2019, rather than penalizing those that need/want to help users that still use SLE_11, for whatever reason. As proposed I oppose the change. Later, Robert -- Robert Schweikert MAY THE SOURCE BE WITH YOU Public Cloud Architect LINUX rjschwei@suse.com IRC: robjo
Am 22.11.2016 um 17:29 schrieb Robert Schweikert:
As proposed I oppose the change.
I oppose the change, too. -- Christian ---------------------------------------------------- - Please do not 'CC' me on list mails. Just reply to the list :) ---------------------------------------------------- http://www.sc24.de - Sportbekleidung ---------------------------------------------------- -- To unsubscribe, e-mail: opensuse-packaging+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-packaging+owner@opensuse.org
On Tue, 2016-11-22 at 17:43 +0100, Christian wrote:
Am 22.11.2016 um 17:29 schrieb Robert Schweikert:
As proposed I oppose the change.
I oppose the change, too.
Can we extend a bit more? As I see in monitoring[1] there are several packages that do not build in sle11 anymore. According to the page there are * succeeded: 1386 * failed: 154 * unresolvable: 175 * broken: 3 * disabled: 239 So a significant portion of the packages are not working. I really doubt that SLE11 users are using d:l:p anymore (but I can be wrong). [1] https://build.opensuse.org/project/monitor/devel:languages:python -- SUSE Linux GmbH, GF: Felix Imendörffer, Jane Smithard, Graham Norton, HRB 21284 (AG Nürnberg) Maxfeldstraße 5, 90409 Nürnberg, Germany -- To unsubscribe, e-mail: opensuse-packaging+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-packaging+owner@opensuse.org
On 11/23/2016 03:47 AM, Alberto Planas Dominguez wrote:
On Tue, 2016-11-22 at 17:43 +0100, Christian wrote:
Am 22.11.2016 um 17:29 schrieb Robert Schweikert:
As proposed I oppose the change.
I oppose the change, too.
Can we extend a bit more? As I see in monitoring[1] there are several packages that do not build in sle11 anymore. According to the page there are
* succeeded: 1386 * failed: 154 * unresolvable: 175 * broken: 3 * disabled: 239
So a significant portion of the packages are not working.
I really doubt that SLE11 users are using d:l:p anymore (but I can be wrong).
[1] https://build.opensuse.org/project/monitor/devel:languages:python
Maybe we can take a half step instead. Keep the repo's enabled but advise maintainers that they are not obliged to keep SLE-11 building if it requires additional effort on there behalf, in that way the people that do care about having package X available on SLE-11 and that are willing to make it work can do so. Packages that are unresolvable, known to have bugs that stop them working in a significant manner or will have security issues under SLE-11 should be disabled for build on SLE-11 with the reason documented somewhere so people know what it takes to get re enabled. We could extend this further to packages not building (when someone branches the package its easy enough to see why it was disabled). These packages could then be reenabled for build if someone steps up to fix them. If we get to a point in the future where say 75% (or any other agreed number) of packages are disabled or failing to build we could then question again whether its worth dropping the repo as a whole. Cheers -- Simon Lees (Simotek) http://simotek.net Emergency Update Team keybase.io/simotek SUSE Linux Adeliade Australia, UTC+9:30 GPG Fingerprint: 5B87 DB9D 88DC F606 E489 CEC5 0922 C246 02F0 014B
On 23.11.2016 01:51, Simon Lees wrote:
On 11/23/2016 03:47 AM, Alberto Planas Dominguez wrote:
On Tue, 2016-11-22 at 17:43 +0100, Christian wrote:
Am 22.11.2016 um 17:29 schrieb Robert Schweikert:
As proposed I oppose the change.
I oppose the change, too.
Can we extend a bit more? As I see in monitoring[1] there are several packages that do not build in sle11 anymore. According to the page there are
* succeeded: 1386 * failed: 154 * unresolvable: 175 * broken: 3 * disabled: 239
So a significant portion of the packages are not working.
I really doubt that SLE11 users are using d:l:p anymore (but I can be wrong).
[1] https://build.opensuse.org/project/monitor/devel:languages:python
Maybe we can take a half step instead.
Keep the repo's enabled but advise maintainers that they are not obliged to keep SLE-11 building if it requires additional effort on there behalf, in that way the people that do care about having package X available on SLE-11 and that are willing to make it work can do so. Packages that are unresolvable, known to have bugs that stop them working in a significant manner or will have security issues under SLE-11 should be disabled for build on SLE-11 with the reason documented somewhere so people know what it takes to get re enabled.
Yes, this should also keep copies of the last successful build in the repo - which is exactly what best serves SLES11 users. SP4 is still in wide usage and support. -- Ralf Lang Linux Consultant / Developer Tel.: +49-170-6381563 Mail: lang@b1-systems.de B1 Systems GmbH Osterfeldstraße 7 / 85088 Vohburg / http://www.b1-systems.de GF: Ralph Dehner / Unternehmenssitz: Vohburg / AG: Ingolstadt,HRB 3537 -- To unsubscribe, e-mail: opensuse-packaging+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-packaging+owner@opensuse.org
On 11/22/2016 05:51 PM, Simon Lees wrote:
Maybe we can take a half step instead.
Keep the repo's enabled but advise maintainers that they are not obliged to keep SLE-11 building if it requires additional effort on there behalf, in that way the people that do care about having package X available on SLE-11 and that are willing to make it work can do so. Packages that are unresolvable, known to have bugs that stop them working in a significant manner or will have security issues under SLE-11 should be disabled for build on SLE-11 with the reason documented somewhere so people know what it takes to get re enabled. We could extend this further to packages not building (when someone branches the package its easy enough to see why it was disabled). These packages could then be reenabled for build if someone steps up to fix them.
If we get to a point in the future where say 75% (or any other agreed number) of packages are disabled or failing to build we could then question again whether its worth dropping the repo as a whole.
Cheers
This sounds extremely reasonable and better to me than actively removing all SLE11 compatibility workarounds, at least at the present time. -- Jason Craig -- To unsubscribe, e-mail: opensuse-packaging+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-packaging+owner@opensuse.org
On 23.11.2016 01:51, Simon Lees wrote:
Maybe we can take a half step instead.
how about a three-quarters-step? Keep the repo and disable building by default. IIUC this should also keep built packages intact, and allow people who use SLE11 to enable support as required. This would also show how many packages the SLE11 people actually care about. regards m.
Keep the repo's enabled but advise maintainers that they are not obliged to keep SLE-11 building if it requires additional effort on there behalf, in that way the people that do care about having package X available on SLE-11 and that are willing to make it work can do so. Packages that are unresolvable, known to have bugs that stop them working in a significant manner or will have security issues under SLE-11 should be disabled for build on SLE-11 with the reason documented somewhere so people know what it takes to get re enabled. We could extend this further to packages not building (when someone branches the package its easy enough to see why it was disabled). These packages could then be reenabled for build if someone steps up to fix them.
If we get to a point in the future where say 75% (or any other agreed number) of packages are disabled or failing to build we could then question again whether its worth dropping the repo as a whole.
Cheers
On Thu, Nov 24, 2016 at 10:11 AM, jan matejek
On 23.11.2016 01:51, Simon Lees wrote:
Maybe we can take a half step instead.
Keep the repo's enabled but advise maintainers that they are not obliged to keep SLE-11 building if it requires additional effort on there behalf, in that way the people that do care about having package X available on SLE-11 and that are willing to make it work can do so. Packages that are unresolvable, known to have bugs that stop them working in a significant manner or will have security issues under SLE-11 should be disabled for build on SLE-11 with the reason documented somewhere so people know what it takes to get re enabled. We could extend this further to packages not building (when someone branches the package its easy enough to see why it was disabled). These packages could then be reenabled for build if someone steps up to fix them.
If we get to a point in the future where say 75% (or any other agreed number) of packages are disabled or failing to build we could then question again whether its worth dropping the repo as a whole.
Cheers
how about a three-quarters-step?
Keep the repo and disable building by default. IIUC this should also keep built packages intact, and allow people who use SLE11 to enable support as required.
This would also show how many packages the SLE11 people actually care about.
regards m.
This sounds more reasonable. But the fact that more than 1/3 of the packages are already either failing, unresolvable, or disabled suggests to me that people don't care that much about SLE-11. And keep in mind the number of failing packages is going to go up once the packages that currently are only present in devel:languages:python3 are merged into devel:languages:python. The question I guess is how easy it will be to handle the lack of python 3 and noarch support in the single spec file system. -- To unsubscribe, e-mail: opensuse-packaging+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-packaging+owner@opensuse.org
On 11/29/2016 06:06 AM, Todd Rme wrote:
On Thu, Nov 24, 2016 at 10:11 AM, jan matejek
wrote: On 23.11.2016 01:51, Simon Lees wrote:
Maybe we can take a half step instead.
Keep the repo's enabled but advise maintainers that they are not obliged to keep SLE-11 building if it requires additional effort on there behalf, in that way the people that do care about having package X available on SLE-11 and that are willing to make it work can do so. Packages that are unresolvable, known to have bugs that stop them working in a significant manner or will have security issues under SLE-11 should be disabled for build on SLE-11 with the reason documented somewhere so people know what it takes to get re enabled. We could extend this further to packages not building (when someone branches the package its easy enough to see why it was disabled). These packages could then be reenabled for build if someone steps up to fix them.
If we get to a point in the future where say 75% (or any other agreed number) of packages are disabled or failing to build we could then question again whether its worth dropping the repo as a whole.
Cheers
how about a three-quarters-step?
Keep the repo and disable building by default. IIUC this should also keep built packages intact, and allow people who use SLE11 to enable support as required.
This would also show how many packages the SLE11 people actually care about.
regards m.
This sounds more reasonable. But the fact that more than 1/3 of the packages are already either failing, unresolvable, or disabled suggests to me that people don't care that much about SLE-11. And keep in mind the number of failing packages is going to go up once the packages that currently are only present in devel:languages:python3 are merged into devel:languages:python.
Yeah I guess it comes down to what those packages are rather then just looking at the numbers, each person (presuming they exist) may only have a few packages that they care about and they are probably keeping these working but don't care as much about the rest of the packages in the repo.
The question I guess is how easy it will be to handle the lack of python 3 and noarch support in the single spec file system.
Yeah I carefully worded what I wrote above to try and say if there is big issues there then it would be good grounds for dropping everything if know one can fix it but I guess we will wait and see. If we did have to drop it we could always create a SLE-11 repo and move everything that builds there before applying the change. -- Simon Lees (Simotek) http://simotek.net Emergency Update Team keybase.io/simotek SUSE Linux Adeliade Australia, UTC+9:30 GPG Fingerprint: 5B87 DB9D 88DC F606 E489 CEC5 0922 C246 02F0 014B
On Tue, Nov 22, 2016 at 7:51 PM, Simon Lees
On 11/23/2016 03:47 AM, Alberto Planas Dominguez wrote:
On Tue, 2016-11-22 at 17:43 +0100, Christian wrote:
Am 22.11.2016 um 17:29 schrieb Robert Schweikert:
As proposed I oppose the change.
I oppose the change, too.
Can we extend a bit more? As I see in monitoring[1] there are several packages that do not build in sle11 anymore. According to the page there are
* succeeded: 1386 * failed: 154 * unresolvable: 175 * broken: 3 * disabled: 239
So a significant portion of the packages are not working.
I really doubt that SLE11 users are using d:l:p anymore (but I can be wrong).
[1] https://build.opensuse.org/project/monitor/devel:languages:python
Maybe we can take a half step instead.
Keep the repo's enabled but advise maintainers that they are not obliged to keep SLE-11 building if it requires additional effort on there behalf, in that way the people that do care about having package X available on SLE-11 and that are willing to make it work can do so. Packages that are unresolvable, known to have bugs that stop them working in a significant manner or will have security issues under SLE-11 should be disabled for build on SLE-11 with the reason documented somewhere so people know what it takes to get re enabled. We could extend this further to packages not building (when someone branches the package its easy enough to see why it was disabled). These packages could then be reenabled for build if someone steps up to fix them.
From what I recall this is already the policy. The problem is that "the people that do care" aren't stepping up to do the work, but failing packages keep getting re-enabled and left failing.
SLE-11 is the only supported platform that doesn't support Python 3 and, more importantly, the only supported platform that doesn't support noarch Python packages. -- To unsubscribe, e-mail: opensuse-packaging+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-packaging+owner@opensuse.org
On mardi, 22 novembre 2016 10.52:37 h CET Todd Rme wrote:
I propose we remove the SLE 11 repository from devel:languages:python. Most spec files require special workaround to build at all with this target and it requires backporting a bunch of system-critical packages that no other release requires, and even then many packages fail to build. And the scheduler punishes the repo for having failures.
The amount of effort needed to keep basic, key dependencies working is growing constantly. And these workarounds means that it will be essentially infeasible to support it in the upcoming single spec file system, which means most key packages will no longer be able to support it anyway.
Anyone who really needs specific packages can always link to them and build them in their home repo.
So I think we should go ahead and remove it. Once it is removed, packages that are updated can have the workarounds deleted
I'm 50% in favor of (ease also the transition to a unified d:l:p) as dlp3 didn't have SLE11 as target. But in the same time, supporting users with SLE11 is also important. Isn't it possible de copy/link dlp for SLE11 in the actual version. Make it build once, and then let a specific repository. (I don't know if those users need a absolute fresh package). Didn't we have a backport repository for SLE11 where we can add those packages ? Then all needed changes can occurs in d:l:p, and people wanting to refresh SLE11 backports would be able to do it on their own repository (I guess with a lot less rebuilds) thoughts? -- Bruno Friedmann Ioda-Net Sàrl www.ioda-net.ch Bareos Partner, openSUSE Member, fsfe fellowship GPG KEY : D5C9B751C4653227 irc: tigerfoot -- To unsubscribe, e-mail: opensuse-packaging+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-packaging+owner@opensuse.org
participants (10)
-
Alberto Planas Dominguez
-
Bruno Friedmann
-
Christian
-
jan matejek
-
Jason Craig
-
Ralf Lang
-
Richard Brown
-
Robert Schweikert
-
Simon Lees
-
Todd Rme