On So, 2008-08-17 at 19:03 +0200, Michael Karcher wrote:
configure | 303 +++++++++++++++++++++++++++++++++++++++++++++++++++
You should never include autogenerated code in a Patch. This will reduce the risk, that your Patch does not apply after a different Patch was commited.
Am Montag, den 18.08.2008, 09:30 +0200 schrieb Detlef Riekenberg:
On So, 2008-08-17 at 19:03 +0200, Michael Karcher wrote:
configure | 303 +++++++++++++++++++++++++++++++++++++++++++++++++++
You should never include autogenerated code in a Patch. This will reduce the risk, that your Patch does not apply after a different Patch was commited.
I feel the same, but I thought that configure is not in the toplevel .gitignore on some purpose I don't understand. If I should not include that file into the patch, shouldn't the generated .gitignore be fixed? (This also applies to include/config.h.in)
Regards, Michael Karcher
"Detlef Riekenberg" wine.dev@web.de wrote:
On So, 2008-08-17 at 19:03 +0200, Michael Karcher wrote:
configure | 303 +++++++++++++++++++++++++++++++++++++++++++++++++++
You should never include autogenerated code in a Patch. This will reduce the risk, that your Patch does not apply after a different Patch was commited.
Btw, how does patchwatcher handle that? Does it detect that in the case of changing config.h.in, dlls/Makefile.in, server/protocol.def it needs to regenerate the files by running autoconf, tools/make_requests, or tools/make_makefiles?
On Monday 18 August 2008 09:50:05 Dmitry Timoshkov wrote:
"Detlef Riekenberg" wine.dev@web.de wrote:
On So, 2008-08-17 at 19:03 +0200, Michael Karcher wrote:
configure | 303 +++++++++++++++++++++++++++++++++++++++++++++++++++
You should never include autogenerated code in a Patch. This will reduce the risk, that your Patch does not apply after a different Patch was commited.
Btw, how does patchwatcher handle that? Does it detect that in the case of changing config.h.in, dlls/Makefile.in, server/protocol.def it needs to regenerate the files by running autoconf, tools/make_requests, or tools/make_makefiles?
Not at all.
Cheers, Kai
On Mon, 18 Aug 2008, Detlef Riekenberg wrote:
On So, 2008-08-17 at 19:03 +0200, Michael Karcher wrote:
configure | 303 +++++++++++++++++++++++++++++++++++++++++++++++++++
You should never include autogenerated code in a Patch.
How do you do that with Git?
When I modify configure.ac I do a local commit of configure.ac and configure, since they are a single changeset. But then when I use git-format-patch to prepare the email to wine-patches, both configure.ac and configure get included in that email.
So do you then edit the email that got prepared by git-format-patch? Do you do two local commits (seems like a pain and ugly)?
Francois Gouget wrote:
On Mon, 18 Aug 2008, Detlef Riekenberg wrote:
On So, 2008-08-17 at 19:03 +0200, Michael Karcher wrote:
configure | 303 +++++++++++++++++++++++++++++++++++++++++++++++++++
You should never include autogenerated code in a Patch.
How do you do that with Git?
When I modify configure.ac I do a local commit of configure.ac and configure, since they are a single changeset. But then when I use
That is not really a single changeset. Not logically (one a real change the other an autogenerated one) nor technically as both files can be modified independently without breaking the compile of Wine.
git-format-patch to prepare the email to wine-patches, both configure.ac and configure get included in that email.
So do you then edit the email that got prepared by git-format-patch? Do you do two local commits (seems like a pain and ugly)?
Two commits is easier and cleaner imho. You don't have to edit the patch before sending and it's easier to rebase on upstream after Alexandre committed your patches. The configure patch you can then just "git rebase --skip". Hmm ... actually you can do the same with the combined patch too. Though the configure patch can have a nice commit message that makes it visually easy to see that you can safely skip the patch without loosing anything. With the combined patch one would have to check that really only the configure part conflicted (Alexandre might have edited the configure.ac part).
bye michael