Can somebody change the appdb to allow submitting results against 0.9.47? I think that's what Gutsy's bundled version of wine is. Somebody just submitted bogus results (with the wrong wine version) because we lack that option.
The appdb code currently prunes the list to several of the most recent versions so people don't submit entries against older or obsolete versions.
Is it really useful to have test results submitted such an old version when we almost always ask users to upgrade to the latest version in the #winehq channel or in bugzilla?
We do mention that if their version is older than the listed version the user should upgrade. Users that submit results with the wrong or old version may mean well but may not be the kind of test results we want to rely on.
Chris
On Sun, Apr 6, 2008 at 2:47 PM, Dan Kegel dank@kegel.com wrote:
Can somebody change the appdb to allow submitting results against 0.9.47? I think that's what Gutsy's bundled version of wine is. Somebody just submitted bogus results (with the wrong wine version) because we lack that option.
On Sun, Apr 6, 2008 at 11:56 AM, Chris Morgan chmorgan@gmail.com wrote:
The appdb code currently prunes the list to several of the most recent versions so people don't submit entries against older or obsolete versions.
Of course. It's just a question of the right cutoff. It seems reasonable to me to include the version bundled with gutsy, even if it's getting old, because gutsy is so popular.
Is it really useful to have test results submitted such an old version when we almost always ask users to upgrade to the latest version in the #winehq channel or in bugzilla?
Perhaps. Some things did run ok with it, and we have had some regressions since then.
We do mention that if their version is older than the listed version the user should upgrade. Users that submit results with the wrong or old version may mean well but may not be the kind of test results we want to rely on.
Sure, but one goes to market with the users one has, not with the users one wishes one had :-) - Dan
Dan Kegel wrote:
On Sun, Apr 6, 2008 at 11:56 AM, Chris Morgan chmorgan@gmail.com wrote:
The appdb code currently prunes the list to several of the most recent versions so people don't submit entries against older or obsolete versions.
Of course. It's just a question of the right cutoff. It seems reasonable to me to include the version bundled with gutsy, even if it's getting old, because gutsy is so popular.
That's the whole point of requiring upgrades. And ubuntu can package a newer version to their distro as well. I don't see not point in supporting this. All Wine versions can and should be considered as bug fixes.
So I really against this. For now we have to force users upgrade if distro doesn't.
Vitaliy.
On Sun, Apr 6, 2008 at 3:02 PM, Dan Kegel dank@kegel.com wrote:
On Sun, Apr 6, 2008 at 11:56 AM, Chris Morgan chmorgan@gmail.com wrote:
The appdb code currently prunes the list to several of the most recent versions so people don't submit entries against older or obsolete versions.
Of course. It's just a question of the right cutoff. It seems reasonable to me to include the version bundled with gutsy, even if it's getting old, because gutsy is so popular.
If we wait another 18 days can we ask users of gutsy to upgrade to hardy? ;-)
If we ask users to put the wine ubuntu repository in their sources.list file won't that let them continue to keep up with newer releases without much difficulty?
Is it really useful to have test results submitted such an old version when we almost always ask users to upgrade to the latest version in the #winehq channel or in bugzilla?
Perhaps. Some things did run ok with it, and we have had some regressions since then.
If we don't plan on releasing new versions based on the release in gutsy, 0.9.47, then I think we've got to consider that version to be obsolete and unsupported. It also doesn't make much sense for us to be storing test results for older versions except so we can perform analysis of the results. We can change the wording to be more explicit about not submitting results from older versions.
We do mention that if their version is older than the listed version the user should upgrade. Users that submit results with the wrong or old version may mean well but may not be the kind of test results we want to rely on.
Sure, but one goes to market with the users one has, not with the users one wishes one had :-)
- Dan
I agree, we can't always choose the kind of data we get. It's up to us to decide which test results to keep and which ones to discard.
Chris
On Sun, Apr 6, 2008 at 1:49 PM, Chris Morgan chmorgan@gmail.com wrote:
It seems reasonable to me to include the version bundled with gutsy, even if it's getting old, because gutsy is so popular.
If we wait another 18 days can we ask users of gutsy to upgrade to hardy? ;-)
Of course.
If we ask users to put the wine ubuntu repository in their sources.list file won't that let them continue to keep up with newer releases without much difficulty?
All of your suggestions are along the lines "change the user's behavior" rather than "adapt winehq to the user's behavior". While all your suggestions are fine, they don't change the fact that users are misbehaving because the appdb doesn't allow reports with Gutsy's version of wine. This problem will keep happening with future releases of Ubuntu if we cut off the version of wine too early, as we do now.
If we don't plan on releasing new versions based on the release in gutsy, 0.9.47, then I think we've got to consider that version to be obsolete and unsupported.
By your logic, we should not allow any reports against anything but the latest version of wine.
I agree, we can't always choose the kind of data we get. It's up to us to decide which test results to keep and which ones to discard.
Whatever. I'm tired of discussing this with you. If the appdb maintainers agree with me, they'll make the change. I don't enough to keep arguing. - Dan
On Sun, Apr 6, 2008 at 5:05 PM, Dan Kegel dank@kegel.com wrote:
On Sun, Apr 6, 2008 at 1:49 PM, Chris Morgan chmorgan@gmail.com wrote:
It seems reasonable to me to include the version bundled with gutsy, even if it's getting old, because gutsy is so popular.
If we wait another 18 days can we ask users of gutsy to upgrade to hardy? ;-)
Of course.
If we ask users to put the wine ubuntu repository in their sources.list file won't that let them continue to keep up with newer releases without much difficulty?
All of your suggestions are along the lines "change the user's behavior" rather than "adapt winehq to the user's behavior". While all your suggestions are fine, they don't change the fact that users are misbehaving because the appdb doesn't allow reports with Gutsy's version of wine.
I'm all for adapting our behavior to better fit with the behavior of our users. I've spent hundreds of hours working on improving the appdb.
If there was a good reason stated to accept results for obsoleted versions then I'd agree with you. In the future I can see a need to pin certain versions to the available versions list, certainly 1.0 seems like a pretty obvious version to pin to the list for a period of time.
There are several issues with keeping older test results. We run the risk of confusing users by showing ratings for Wine releases so old that the current behavior doesn't match. We can hide these older versions but why accept them for submission then? If we show them then we need to consider how to present the ratings so users of newer versions don't get confused with older versions that come with some distributions.
By accepting results for older versions we may not be making the user fully aware that their version is considered to be old.
It's a complex issue that deserves more discussion and may require more php and ui work to present properly to the user.
This problem will keep happening with future releases of Ubuntu if we cut off the version of wine too early, as we do now.
I'm a fan of ubuntu, run it on all of my machines etc but that doesn't automatically mean that test results for older versions are useful. We are rapidly approaching a 1.0 release and each incremental release has a wide range of improvements. A user of Gutsy may find it useful to see test results for 0.9.47 but its likely even more useful for us to help them upgrade to a version we support, currently the latest incremental release.
If we don't plan on releasing new versions based on the release in gutsy, 0.9.47, then I think we've got to consider that version to be obsolete and unsupported.
By your logic, we should not allow any reports against anything but the latest version of wine.
The reason I kept the last several entries on the list was because some people tend to lag behind in their wine versions. Its understandable for us to have a window of versions so we can get reasonably current test results, even if not the most recent test results. Already the list contains a handful of previous releases, several months of obsolete versions.
I agree, we can't always choose the kind of data we get. It's up to us to decide which test results to keep and which ones to discard.
Whatever. I'm tired of discussing this with you. If the appdb maintainers agree with me, they'll make the change. I don't enough to keep arguing.
- Dan
I'm not opposed to a patch that re-adds 0.9.47 version to the list but I'd like to see a patch that adds a table to keep pinned versions and php code to manage this list.
Chris
On Sun, Apr 6, 2008 at 4:55 PM, Chris Morgan chmorgan@gmail.com wrote:
On Sun, Apr 6, 2008 at 5:05 PM, Dan Kegel dank@kegel.com wrote:
On Sun, Apr 6, 2008 at 1:49 PM, Chris Morgan chmorgan@gmail.com wrote:
It seems reasonable to me to include the version bundled with gutsy, even if it's getting old, because gutsy is so popular.
Has anyone considering talking to Canonical about getting a more recent wine version included in wine? While they understandably don't want to introduce too new of versions, they also need to take into account that we release every two weeks, much faster than most other projects.
On Monday 07 April 2008 04:14:59 Austin English wrote:
Has anyone considering talking to Canonical about getting a more recent wine version included in wine? While they understandably don't want to introduce too new of versions, they also need to take into account that we release every two weeks, much faster than most other projects.
Not going to work. The whole point of a distribution release is to keep the versions of the programs stable and just do security fixes. That way, you're safe from any potential regression in the programs you're using. I agree that Wine tends to release too often for the usual distro release cycle, but I'd argue that so far we've not done a too good job about _not_ having regressions. So I doubt the distros will upgrade the Wine versions post-release.
Cheers, Kai
On Sun, 6 Apr 2008, Chris Morgan wrote: [...]
I'm all for adapting our behavior to better fit with the behavior of our users. I've spent hundreds of hours working on improving the appdb.
If there was a good reason stated to accept results for obsoleted versions then I'd agree with you. In the future I can see a need to pin certain versions to the available versions list, certainly 1.0 seems like a pretty obvious version to pin to the list for a period of time.
If the cutoff is '0.9.48' (for instance), then maybe we could add a 'pre-0.9.48' version. That way users looking for '0.9.47' would know what to use. This would also let bugzilla maintainers easily spot bugs against obsolete versions.
Then when the cutoff evolves to 0.9.53 or whatever we would rename the 'pre-0.9.48' version to 'pre-0.9.53', etc. If renaming versions is not allowed in Bugzilla, then we'd need a more stable name for the catch all version, maybe something like 'older versions' or ome such.
The one remaining problem may be to get the version order-right but that can probably be solved by getting creative with the version name, e.g. '0.9.47.or.older'.
On Mon, Apr 7, 2008 at 2:30 AM, Francois Gouget fgouget@free.fr wrote:
If the cutoff is '0.9.48' (for instance), then maybe we could add a 'pre-0.9.48' version.
Sounds too confusing to me, offhand. I'd rather stick with the current situation of having a too-early cutoff.
That way users looking for '0.9.47' would know what to use. This would also let bugzilla maintainers easily spot bugs against obsolete versions.
You mean appdb... - Dan
Chris Morgan wrote:
On Sun, Apr 6, 2008 at 5:05 PM, Dan Kegel dank@kegel.com wrote:
On Sun, Apr 6, 2008 at 1:49 PM, Chris Morgan chmorgan@gmail.com wrote:
It seems reasonable to me to include the version bundled with gutsy, even if it's getting old, because gutsy is so popular.
If we wait another 18 days can we ask users of gutsy to upgrade to hardy? ;-)
Of course.
If we ask users to put the wine ubuntu repository in their sources.list file won't that let them continue to keep up with newer releases without much difficulty?
All of your suggestions are along the lines "change the user's behavior" rather than "adapt winehq to the user's behavior". While all your suggestions are fine, they don't change the fact that users are misbehaving because the appdb doesn't allow reports with Gutsy's version of wine.
I'm all for adapting our behavior to better fit with the behavior of our users. I've spent hundreds of hours working on improving the appdb.
If there was a good reason stated to accept results for obsoleted versions then I'd agree with you. In the future I can see a need to pin certain versions to the available versions list, certainly 1.0 seems like a pretty obvious version to pin to the list for a period of time.
There are several issues with keeping older test results. We run the risk of confusing users by showing ratings for Wine releases so old that the current behavior doesn't match. We can hide these older versions but why accept them for submission then? If we show them then we need to consider how to present the ratings so users of newer versions don't get confused with older versions that come with some distributions.
My policy for Ubuntu is this: there should be no regressions from the Wine included in one Ubuntu release to the next. That means, for Hardy, no regressions from 0.9.47 to whatever I put in Hardy (which will probably be 0.9.59 after a patch or two).
It's a lot easier for me to do this if we don't prohibit test data for that version.
I'm a fan of ubuntu, run it on all of my machines etc but that doesn't automatically mean that test results for older versions are useful. We are rapidly approaching a 1.0 release and each incremental release has a wide range of improvements. A user of Gutsy may find it useful to see test results for 0.9.47 but its likely even more useful for us to help them upgrade to a version we support, currently the latest incremental release.
If their applications are working, then users shouldn't be using the apt repository and latest Wine version because it might introduce a regression. We are, in a very real sense, making AppDB less useful for them by not including the results.
I'm not opposed to a patch that re-adds 0.9.47 version to the list but I'd like to see a patch that adds a table to keep pinned versions and php code to manage this list.
It's not that big a deal now, so long as the Hardy version (0.9.59 likely) stays around for a while ;)
Thanks, Scott Ritchie
On Mon, Apr 7, 2008 at 11:40 PM, Scott Ritchie scott@open-vote.org wrote:
It's not that big a deal now, so long as the Hardy version (0.9.59 likely) stays around for a while ;)
I think Chris was planning on removing 0.9.59 from the list way before Hardy is unsupported. Chris?
On Tue, Apr 8, 2008 at 9:30 AM, Dan Kegel dank@kegel.com wrote:
On Mon, Apr 7, 2008 at 11:40 PM, Scott Ritchie scott@open-vote.org wrote:
It's not that big a deal now, so long as the Hardy version (0.9.59 likely) stays around for a while ;)
I think Chris was planning on removing 0.9.59 from the list way before Hardy is unsupported. Chris?
Will Hardy upgrade to Wine 1.0 when it is released? I don't believe anyone is opposed to keeping particular important releases on the list but do we do that for distributions upon request? Only ones with lots of users of wine? The question I had was what results from an older version buy us or other users. If users can easily upgrade via using the winehq source in Ubuntu's then maybe it isn't as big of a deal. The reason for pruning the list was to avoid results from old versions and to simplify the test submission process by reducing the number of entries in that dropdown.
It seems like we need to come up with a criteria to pin versions.
I'll submit a bug for it on bugzilla today so we don't forget that we need to add that feature with the goal of having it before 1.0 is shipped but maybe we'll have it sooner.
Chris
-
Austin English -
Chris Morgan -
Dan Kegel -
Francois Gouget -
Kai Blin -
Scott Ritchie -
Vitaliy Margolen