Re: [sage-devel] check merge failure on Trac

['Reimundo Heluani' via sage-devel]

On August 28, 2020 10:00:49 AM GMT-03:00, Michael Orlitzky 
<mich...@orlitzky.com> wrote:
>On 2020-08-28 08:40, 'Reimundo Heluani' via sage-devel wrote:
>> On Aug 28, Michael Orlitzky wrote:
>>> On 2020-08-28 08:23, 'Reimundo Heluani' via sage-devel wrote:
>>>> Thanks, I guess recompilation is unavoidable then, the worst part
>is that it was one of my own tickets :)
>>>>
>>>
>>> Dima meant "git fetch" literally. That will pull down the changes,
>but
>>> won't try to merge them into your local branch (like "git pull"
>would).
>>>
>> yeah but looking at a diff would be impossible to find conflicts with
>such a 
>> large patch. The only way I know of is attempting a merge, and if
>there are 
>> conflicts this always results in changing the working tree. 
>
>You can try creating a patch with `git format-patch`, and then applying
>it with `git apply --check`.
>
>I have a branch of the ClamAV project right now that conflicts with
>upstream, so I'm using that as my example. First I have to find the
>latest common ancestor of my branch and the upstream one,
>
>  $ git merge-base HEAD upstream/dev/0.103
>  854d38de546a53e3c27b90965b8d89804acfdf02
>
>Now if I format-patch that against the upstream branch, I get...
>
>  $ git format-patch 854d38de546..upstream/dev/0.103
>  0000-cover-letter.patch
>  0001-Add-newlines-at-end-of-sources-to-satisfy-strict-war.patch
>  0002-Update-acknowledgements-in-the-NEWS.patch
>  0003-Remove-v-typo-from-unit-test-source.patch
>  0004-bb12596-Fix-build-issue-when-libcheck-missing.patch
>  0005-Add-autogen.sh-to-Autotools-dist.patch
>  0006-Autotools-freshclam-remove-BUILD_CLAMD-macro.patch
>  0007-CMake-Fix-libclamav-pcre2-include-path-issue.patch
>  0008-clam-d-scan-win32-fix-RO-directory-remove-bug.patch
>  0009-Autotools-Add-pkg-config-support-for-finding-pcre2.patch
>  0010-Update-acknowledgements-in-the-NEWS.patch
>
>And one of those should fail to apply:
>
>  $ git apply --check 0001*.patch
>  $ git apply --check 0002*.patch
>  $ git apply --check 0003*.patch
>  $ git apply --check 0004*.patch
>  $ git apply --check 0005*.patch
>  $ git apply --check 0006*.patch
>  error: patch failed: clamav-milter/Makefile.am:36
>  error: clamav-milter/Makefile.am: patch does not apply
>  error: patch failed: clamd/Makefile.am:46
>  error: clamd/Makefile.am: patch does not apply
>  error: patch failed: etc/Makefile.am:23
>  error: etc/Makefile.am: patch does not apply
>
>That last one is responsible for the conflict, and `git status` shows
>that my local tree hasn't been changed (aside from the new patch
>files).
This is very good to know thanks, I'll try this. But I think in general the 
problem that I want to avoid would not be seen by simply a git status. Since

$git checkout develop
$git checkout mybranch

Will trigger recompilation because of the timestamp changes. And git status 
would show everything clean. If the format patch check does not change 
timestamps that's exactly what I was looking for. Thanks

R
-- 
Sent from my Android device with K-9 Mail. Please excuse my brevity.

-- 
You received this message because you are subscribed to the Google Groups 
"sage-devel" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to sage-devel+unsubscr...@googlegroups.com.
To view this discussion on the web visit 
https://groups.google.com/d/msgid/sage-devel/E91F7307-E2B3-47AC-815C-729B0BDA51C3%40potuz.net.