Its Alive! Patchwatcher is reborn.
All, As you may have noticed in the last WWN I have been working on getting patchwatcher back up on a rather nice server whose cpu time is graciously donated by STWing (stwing.org). I've got it running at the moment at http://winepatch.stwing.upenn.edu/results2/ There are still many, many kinks to be worked out before I would use this as reliable and/or before I even think about enabling sending to patch authors. But if you're curious thats where the results are going and will likely be in the future. -Zach p.s. Kudos to Dan Kegel for a great walkthrough in the PW readme.txt, and of course for putting patchwatcher together in the first place!
On Sunday 31 May 2009 09:05:36 Zachary Goldberg wrote:
All,
As you may have noticed in the last WWN I have been working on getting patchwatcher back up on a rather nice server whose cpu time is graciously donated by STWing (stwing.org). I've got it running at the moment at
The most notable issue is that most patches fail to apply and apply claims they're reversed patches. Are you sure you're using git am the right way? Cheers, Kai -- Kai Blin WorldForge developer http://www.worldforge.org/ Wine developer http://wiki.winehq.org/KaiBlin Samba team member http://www.samba.org/samba/team/ -- Will code for cotton.
On Sun, May 31, 2009 at 3:27 AM, Kai Blin <kai.blin(a)gmail.com> wrote:
On Sunday 31 May 2009 09:05:36 Zachary Goldberg wrote:
All,
As you may have noticed in the last WWN I have been working on getting patchwatcher back up on a rather nice server whose cpu time is graciously donated by STWing (stwing.org). I've got it running at the moment at
The most notable issue is that most patches fail to apply and apply claims they're reversed patches. Are you sure you're using git am the right way?
Cheers, Kai
-- Kai Blin WorldForge developer http://www.worldforge.org/ Wine developer http://wiki.winehq.org/KaiBlin Samba team member http://www.samba.org/samba/team/ -- Will code for cotton.
The code wasn't correctly checking for previously applied patches. Does it look more reasonable now in the number of patch application failures? -Zach
On Sun, May 31, 2009 at 12:05 AM, Zachary Goldberg <zgold(a)bluesata.com> wrote:
All,
As you may have noticed in the last WWN I have been working on getting patchwatcher back up on a rather nice server whose cpu time is graciously donated by STWing (stwing.org). I've got it running at the moment at
http://winepatch.stwing.upenn.edu/results2/
There are still many, many kinks to be worked out before I would use this as reliable and/or before I even think about enabling sending to patch authors. But if you're curious thats where the results are going and will likely be in the future.
-Zach
p.s. Kudos to Dan Kegel for a great walkthrough in the PW readme.txt, and of course for putting patchwatcher together in the first place!
Speaking of pre-commit checking, the Wine wiki page for patchwatcher page [1] currently list the following todo items: Check for C++ comments. Check for nameless unions (GCC 2.95). Check for missing "\n"s in traces/fixmes/etc. Make sure sizeof is not used in traces. These sound like things we can catch with a git pre-commit hook, so developers who use git can fix these even before it hits patchwatcher. [1] http://wiki.winehq.org/PatchWatcher
On Sun, 31 May 2009, Lei Zhang wrote: [...]
Speaking of pre-commit checking, the Wine wiki page for patchwatcher page [1] currently list the following todo items:
Check for C++ comments.
Currently we don't have a good way of detecting these. winapi_check attempts to detect them but it gets fooled by stuff like foo("http://www.winehq.org") or /* http://www.winehq.org */. With a better C parser it may work.
Check for nameless unions (GCC 2.95).
The way I check for these is to compile with gcc 2.95. However nowadays that requires using a Debian 4.0 chroot because gcc 2.95 is not supported anymore by the recent glibc and other libraries. So this is not a good way to detect these issues in a commit hook. I'm not sure detecting such issues would be easy in a script.
Check for missing "\n"s in traces/fixmes/etc.
These have a bunch of exceptions. There are really a bunch of cases where there is intentionally no trailing '\n'. Typically because we're dumping a bunch of flags or the contents of an array in hexadecimal. So the pre commit hook would have to be overridable and would need a regularly updated list of exceptions so people committing stuff on these files don't get warned again and again.
Make sure sizeof is not used in traces.
This looks like it could be reasonably easy to check for. Actually it's not just traces it's all forms of printf. -- Francois Gouget <fgouget(a)free.fr> http://fgouget.free.fr/ "Only wimps use tape backup: _real_ men just upload their important stuff on ftp, and let the rest of the world mirror it ;)" -- Linus Torvalds
participants (4)
-
Francois Gouget -
Kai Blin -
Lei Zhang -
Zachary Goldberg