Converting the wpkg scripts to appinstall doesn't seem like that difficult of a task, mostly just time consuming.
It's harder than it sounds to actually get a few useful app tests written. Plus there'll be some work writing up and fixing the bugs it uncovers.
That is why I thinking working on tying it into test.winehq.org is more worthy. I still think tying appinstall and patchwatcher together would be beneficial and allow people submitting patches to see where regressions pop up in software, and to see what new software works with each submitted patch. Perhaps a project of implementing all of this and adding wpkg scripts would be accepted?
Arjun Comar is going to propose something similar, but using Buildbot as the point of integration. - Dan
On Tue, Mar 9, 2010 at 5:48 AM, Dan Kegel dank@kegel.com wrote:
Converting the wpkg scripts to appinstall doesn't seem like that difficult of a task, mostly just time consuming.
It's harder than it sounds to actually get a few useful app tests written. Plus there'll be some work writing up and fixing the bugs it uncovers.
So then do you think it's a worthy proposal to just work on wpkg scripts? I have no problem working on this, and writting up/trying to fix bugs that are uncovered. I just didn't think that by itself it would be good enough to be accepted as a GSoC project.
That is why I thinking working on tying it into test.winehq.org is more
worthy.
I still think tying appinstall and patchwatcher together would be beneficial and allow people submitting patches to see where regressions pop up in software, and to see what new software works with each submitted patch. Perhaps a project of implementing all of this and adding wpkg scripts
would be accepted?
Arjun Comar is going to propose something similar, but using Buildbot as the point of integration.
- Dan
I look forward to seeing his proposal. Will he be bring patchwatcher back to life, but in buildbot form?
-Seth Shelnutt
On Tue, Mar 9, 2010 at 6:53 AM, Seth Shelnutt shelnutt2@gmail.com wrote:
It's harder than it sounds to actually get a few useful app tests written. Plus there'll be some work writing up and fixing the bugs it uncovers.
So then do you think it's a worthy proposal to just work on wpkg scripts? I have no problem working on this, and writting up/trying to fix bugs that are uncovered. I just didn't think that by itself it would be good enough to be accepted as a GSoC project.
I think it's enough, especially if you do one or two ahead of time and demonstrate that the process actually works. It helps to have a wine patch under your belt, too.
Arjun Comar is going to propose something similar, but using Buildbot as the point of integration.
I look forward to seeing his proposal. Will he be bring patchwatcher back to life, but in buildbot form?
I think he's starting with just getting a buildbot working, then adding a trybot with a manual upload. Patchwatcher integration would be a plus, but its mailing list watcher is fragile and needs rewriting. (Maybe you could do that little bit.)
Arjun, are you ready to post a proposal? - Dan
Yea, I'm just about ready. I was going to wait until I had the initial buildbot system ready to go, but there's no reason I couldn't but my proposal on wine-devel tonight. Quick summary of my proposal: - Buildbot system to automate testing of compilation and application installation / execution (integrating AppInstall) - Several levels of testing depending on time constraints - Easily configurable - Trybot scheduler (manual upload initially, perhaps if as Dan said, patchwatcher is in a good state it could be integrated) for patch testing.
I'll send an email within the next 12 hours or so with my current plan on accomplishing those tasks.
On Tue, Mar 9, 2010 at 10:02 AM, Dan Kegel dank@kegel.com wrote:
On Tue, Mar 9, 2010 at 6:53 AM, Seth Shelnutt shelnutt2@gmail.com wrote:
It's harder than it sounds to actually get a few useful app tests written. Plus there'll be some work writing up and fixing the bugs it uncovers.
So then do you think it's a worthy proposal to just work on wpkg scripts?
I
have no problem working on this, and writting up/trying to fix bugs that
are
uncovered. I just didn't think that by itself it would be good enough to
be
accepted as a GSoC project.
I think it's enough, especially if you do one or two ahead of time and demonstrate that the process actually works. It helps to have a wine patch under your belt, too.
Arjun Comar is going to propose something similar, but using Buildbot as the point of integration.
I look forward to seeing his proposal. Will he be bring patchwatcher back
to
life, but in buildbot form?
I think he's starting with just getting a buildbot working, then adding a trybot with a manual upload. Patchwatcher integration would be a plus, but its mailing list watcher is fragile and needs rewriting. (Maybe you could do that little bit.)
Arjun, are you ready to post a proposal?
- Dan
On Tue, Mar 9, 2010 at 9:19 AM, Arjun Comar mandaya@rose-hulman.edu wrote:
Yea, I'm just about ready. I was going to wait until I had the initial buildbot system ready to go, but there's no reason I couldn't but my proposal on wine-devel tonight. Quick summary of my proposal:
- Buildbot system to automate testing of compilation and application
installation / execution (integrating AppInstall)
- Several levels of testing depending on time constraints
- Easily configurable
- Trybot scheduler (manual upload initially, perhaps if as Dan said,
patchwatcher is in a good state it could be integrated) for patch testing.
I'll send an email within the next 12 hours or so with my current plan on accomplishing those tasks.
Something to keep in mind is that Appinstall takes a while to run...usually at least an hour.
Yea, I noticed. That's why I was planning to include various build parameters based on time constraints.
On Tue, Mar 9, 2010 at 11:31 AM, Austin English austinenglish@gmail.comwrote:
On Tue, Mar 9, 2010 at 9:19 AM, Arjun Comar mandaya@rose-hulman.edu wrote:
Yea, I'm just about ready. I was going to wait until I had the initial buildbot system ready to go, but there's no reason I couldn't but my proposal on wine-devel tonight. Quick summary of my proposal:
- Buildbot system to automate testing of compilation and application
installation / execution (integrating AppInstall)
- Several levels of testing depending on time constraints
- Easily configurable
- Trybot scheduler (manual upload initially, perhaps if as Dan said,
patchwatcher is in a good state it could be integrated) for patch
testing.
I'll send an email within the next 12 hours or so with my current plan on accomplishing those tasks.
Something to keep in mind is that Appinstall takes a while to run...usually at least an hour.
-- -Austin
Alright after creating a few test scripts and playing around with things I think I have finally settled on exactly what my proposal is. It is in three parts,
1) Convert wpkg (or appDeploy) scripts into ahk scripts. I believe that I can convert a hundred or so without to much work.
2) I'd like to implement a more aggressive error reporting. I want to have appinstall rerun failed test using WINEDEBUG=+all (or possibly have it guess what channel is needed to be watched). Saving the output of this will allow one to more easily debug the program. You won't have to go back and rerun the program yourself in wine, you will already be given the output so you can see what failed, where.
3) I'd also like to implement a feature to have the output more in the format of test.winehq.org. I am thinking that it would be easier to have it done split into two scripts, one on client-side, one server-side. Have the client format all the information into an xml file, and then server-side convert that xml into something along the lines of what test.winehq.orglooks like.
I can go more in depth with each step I'd like to take, but I am not sure how specific I should be with my proposal? I was looking at last year's proposals and they didn't seem overly in depth on the steps they planned to take.
The five sample scripts and the edited appinstall.sh can be found here, http://shelnutt2.host56.com/appinstall/ . I've found the easiest way to add the scripts is to copy them into the ~/.appinstallcache folder after it's created and before the tests are run.
Thanks,
-Seth Shelnutt
On Sun, Mar 28, 2010 at 12:53 PM, Seth Shelnutt shelnutt2@gmail.com wrote:
Alright after creating a few test scripts and playing around with things I think I have finally settled on exactly what my proposal is. It is in three parts,
- Convert wpkg (or appDeploy) scripts into ahk scripts. I believe that I
can convert a hundred or so without to much work.
- I'd like to implement a more aggressive error reporting. I want to have
appinstall rerun failed test using WINEDEBUG=+all (or possibly have it guess what channel is needed to be watched). Saving the output of this will allow one to more easily debug the program. You won't have to go back and rerun the program yourself in wine, you will already be given the output so you can see what failed, where.
- I'd also like to implement a feature to have the output more in the
format of test.winehq.org. I am thinking that it would be easier to have it done split into two scripts, one on client-side, one server-side. Have the client format all the information into an xml file, and then server-side convert that xml into something along the lines of what test.winehq.org looks like.
I can go more in depth with each step I'd like to take, but I am not sure how specific I should be with my proposal? I was looking at last year's proposals and they didn't seem overly in depth on the steps they planned to take.
The five sample scripts and the edited appinstall.sh can be found here, http://shelnutt2.host56.com/appinstall/ . I've found the easiest way to add the scripts is to copy them into the ~/.appinstallcache folder after it's created and before the tests are run.
You'll want to make a formal proposal on http://socghop.appspot.com/gsoc/.
One other thing you may want to aim for is better windows support. It should be possible to run appinstall in windows under cygwin (it should be pretty easy to fix, actually).