Closed tueda closed 4 months ago
Hi Takahiro,
At the moment I dare not try this out. First of all I am extremely pressed to have everything ready for tomorrow and second of all I cannot afford that things would not work properly. Hopefully I have some time next week when I will be in Amsterdam and preparing for the trip to Japan.
I can search for regressions ‘manually’. I have done that before. Diff is very useful for that. It is labor intensive, but at least it would not miss anything. Of course that would mess up the history and I do not have the expertise with git to track each contribution. Hence we have to decide what is more important. If we can track the contributors, we can ask them to make the contribution again of course. If they are willing. I will think about it the coming days and also ask Ben. He should be on his way at the moment.
I have to run now to a meeting.
Cheers
Jos
On 11 Apr 2023, at 13:56, Takahiro Ueda @.***> wrote:
So, I made a script https://gist.github.com/tueda/62efdd8524e57c91cd8c770dfe98c9dd that tries to run git cherry-pick https://git-scm.com/docs/git-cherry-pick for all patches in a range of commits. Conflicted patches are skipped.
mkdir tmp cd tmp git clone https://github.com/vermaseren/form.git cd form git checkout v5.0.0-beta.1 wget https://gist.githubusercontent.com/tueda/62efdd8524e57c91cd8c770dfe98c9dd/raw/auto-cherry-pick.sh chmod +x auto-cherry-pick.sh ./auto-cherry-pick.sh FORM-3-1..v5.0.0-beta.1 Then
git log --oneline v5.0.0-beta.1.. or
git log --oneline v5.0.0-beta.1.. sources # only inside the source directory gives candidates of regressions that can be fixed simply by git cherry-pick. Of course, some of them are false positives. And there are some undetected ones (because git cherry-pick conflicts).
Question: how to proceed?
git log --oneline v5.0.0-beta.1.. git log --oneline v5.0.0-beta.1.. sources — Reply to this email directly, view it on GitHub https://github.com/vermaseren/form/issues/441, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABJPCEW64RABRPFJGOVS7KDXAVBITANCNFSM6AAAAAAW2FMM6U. You are receiving this because you are subscribed to this thread.
./check/check.rb --stat valgrind sources/vorm
gives the following failures:
Diagrams_1 (examples.frm:1827) FAILED | | 00:00:00.907
Diagrams_2 (examples.frm:1843) FAILED | | 00:00:00.912
Diagrams_3 (examples.frm:1873) FAILED | | 00:00:00.928
Diagrams_4 (examples.frm:1901) FAILED | | 00:00:01.970
moebius_3 (features.frm:162) FAILED | | 00:00:00.963
Format_noreset_linelen (features.frm:458) FAILED | | 00:00:00.934
Issue392_ContinuationLines_0 (features.frm:1046) FAILED | | 00:00:00.955
Issue7_1 (fixes.frm:119) TIMEOUT |################|>00:05:00.000
Issue7_2 (fixes.frm:135) FAILED | | 00:00:01.127
Issue7_3 (fixes.frm:152) TIMEOUT |################|>00:05:00.000
Issue324 (fixes.frm:2115) FAILED | | 00:00:00.909
Issue325 (fixes.frm:2124) FAILED | | 00:00:00.930
Issue405 (fixes.frm:2278) FAILED | | 00:00:00.911
Issue434 (fixes.frm:2288) FAILED | | 00:00:00.915
Cherry-picking the following commits
commit | subject | fixes | note |
---|---|---|---|
2b2592c | Allow setting ContinuationLines to 0, meaning infinity (#392) | Issue392_ContinuationLines_0 |
|
dd1a71c | Do not reset line length when calling Format with arguments | Format_noreset_linelen |
conflict |
982111a | Fix issue with implicitly or autodeclared dimension of autodeclared index (#324, #325, #405) | Issue324 , Issue325 , Issue405 |
|
bfc7e33 | fix some typos in sources | conflict | |
fb26f35 | fixing another bunch of typos in sources/ | ||
9d7eeb9 | fixing a few typos in sources folder | conflict | |
28e15ea | fix: moebius_ function (#422, #430) | moebius_3 |
|
e7c52d3 | docs: remove invalid licence comment in extcmd.c (#431) | ||
741861a | fix: possible memory error in calling macros with arguments (#434) | Issue434 |
fixes many of them.
Would you guys like me to push these patches? (Or another way to proceed?)
How can we repair "Diagrams" and Issue #7?
Hi Takahiro,
I would say: go ahead. I will be out of touch for more than 3 weeks.
On 19 Apr 2023, at 14:20, Takahiro Ueda @.***> wrote:
./check/check.rb --stat valgrind sources/vorm gives the following failures:
Diagrams_1 (examples.frm:1827) FAILED | | 00:00:00.907 Diagrams_2 (examples.frm:1843) FAILED | | 00:00:00.912 Diagrams_3 (examples.frm:1873) FAILED | | 00:00:00.928 Diagrams_4 (examples.frm:1901) FAILED | | 00:00:01.970 moebius_3 (features.frm:162) FAILED | | 00:00:00.963 Format_noreset_linelen (features.frm:458) FAILED | | 00:00:00.934 Issue392_ContinuationLines_0 (features.frm:1046) FAILED | | 00:00:00.955 Issue7_1 (fixes.frm:119) TIMEOUT |################|>00:05:00.000 Issue7_2 (fixes.frm:135) FAILED | | 00:00:01.127 Issue7_3 (fixes.frm:152) TIMEOUT |################|>00:05:00.000 Issue324 (fixes.frm:2115) FAILED | | 00:00:00.909 Issue325 (fixes.frm:2124) FAILED | | 00:00:00.930 Issue405 (fixes.frm:2278) FAILED | | 00:00:00.911 Issue434 (fixes.frm:2288) FAILED | | 00:00:00.915 Cherry-picking the following commits
commit subject fixes note 2b2592c https://github.com/vermaseren/form/commit/2b2592c63c256077fe464b6c943d90387924ca5f Allow setting ContinuationLines to 0, meaning infinity (#392 https://github.com/vermaseren/form/issues/392) Issue392_ContinuationLines_0
dd1a71c https://github.com/vermaseren/form/commit/dd1a71ce1dfceacdd4e4279c08bb32df84a6c759 Do not reset line length when calling Format with arguments Format_noresetlinelen conflict 982111a https://github.com/vermaseren/form/commit/982111ae7f7cfcb83378748786286f3aea905803 Fix issue with implicitly or autodeclared dimension of autodeclared index (#324 https://github.com/vermaseren/form/issues/324, #325 https://github.com/vermaseren/form/issues/325, #405 https://github.com/vermaseren/form/issues/405) Issue324, Issue325, Issue405
bfc7e33 https://github.com/vermaseren/form/commit/bfc7e33e7aa9af6c95afd53b0a9d777e48242fde fix some typos in sources conflict fb26f35 https://github.com/vermaseren/form/commit/fb26f359033b87858024788e43df3ba9fb8f91ac fixing another bunch of typos in sources/
9d7eeb9 https://github.com/vermaseren/form/commit/9d7eeb9887f327fc7678fbeeda04a79c1452fc26 fixing a few typos in sources folder conflict 28e15ea https://github.com/vermaseren/form/commit/28e15eaf0856a0a012795298d6a4b570e764a8b1 fix: moebius function (#422 https://github.com/vermaseren/form/issues/422, #430 https://github.com/vermaseren/form/issues/430) moebius_3
e7c52d3 https://github.com/vermaseren/form/commit/e7c52d3b07abe21f21718f5e70ee138e856f15ac docs: remove invalid licence comment in extcmd.c (#431 https://github.com/vermaseren/form/issues/431)
741861a https://github.com/vermaseren/form/commit/741861aef8c7fa81bfddf4eab67f3f2cf4ccf53b fix: possible memory error in calling macros with arguments (#434 https://github.com/vermaseren/form/issues/434) Issue434
fixes many of them.Patches (for "git am") Would you guys like me to push these patches? (Or another way to proceed?)
How can we repair "Diagrams" and Issue #7 https://github.com/vermaseren/form/issues/7?
— Reply to this email directly, view it on GitHub https://github.com/vermaseren/form/issues/441#issuecomment-1514636903, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABJPCEWNXJBDN2BLYTLOE6TXB7J77ANCNFSM6AAAAAAW2FMM6U. You are receiving this because you commented.
OK, I will push the above patches. At the moment, I have no idea why Issue #7 was re-broken...
Closing as this has been superseded by #475.
So, I made a script that tries to run
git cherry-pick
for all patches in a range of commits. Conflicted patches are skipped.Then
or
gives candidates of regressions that can be fixed simply by
git cherry-pick
. Of course, some of them are false positives. And there are some undetected ones (becausegit cherry-pick
conflicts).Question: how to proceed?
git log --oneline v5.0.0-beta.1..
``` 14866d8 fix: possible memory error in calling macros with arguments (#434) a5beea4 docs: remove invalid licence comment in extcmd.c (#431) 3188fbd fix: moebius_ function (#422, #430) 8165e80 fixing a few typos in sources folder 658bb45 fixing another bunch of typos in sources/ 7440741 Fix issue with implicitly or autodeclared dimension of autodeclared index cf39bb0 Allow setting ContinuationLines to 0, meaning infinity e5a9db9 Remove more gcc warnings (gcc version dependent) ad7be37 Put in the fix and undid another 3a6b7ec Undoes some of the previous commit to test the failing checks. e3f63b5 Repaired an inefficiency when many expressions are defined (TYPEEXPRESSION) b021935 Repairs the 'famous' GZIP bug. 8917fae Repair of the infamous gzip memory leak. ba2f606 Fixes #219 09b1f39 And repaired the problem with the check in the fix of #55 60af3fd Fixes #55 PrepPoly did not convert some arguments to fast notation. 56295bd Fix for issue #186 ed4e066 Repaired fix.frm 50be42c fixed up the previous fixes and some asserts in the ruby file b7ba717 And yet another bug in TakeContent 91c8499 Fixed gcd_. Was problem with constant terms. f479692 Repaired a sortbot problem with big terms f5a216c Bucket fix and temporary collect % 7d8af84 Partially fix "make check" 415c646 Wrong variable was set to -1 on compiler error 6e8bd2f removed debug output f6973bd small fix in polynomial multiplication 7fd8d58 Changed the PUTNUMBER macro's to avoid warnings on 32-bits systems. 1915da7 fixed small memory leak in poly_to_argument 39b2ebd bugfix: if already factorized, don't factorize again 8661eeb ParFORM: fixed a bug of #write dbecf7f ParFORM: blocked statistics report from slaves c8776ed Little bugfix in %X option in #write. Counting problem. 8174a5a Added tests for examples in the user manual (up to chapter 6 right now). 3189f17 Fixed a bug when testing TFORM. In the multithreaded case the PID was not handled correctly. bb0fe0f Added timeout option. Fixed piped IO to executable (stderr was lost). c2d530f Improved ruby code for testing and added two more test cases. d013702 Initial commit of automatic test suite. c078d4c Add autoconf file. 29fe836 Removed misspelled file. ea97300 Added autoconf files. e5c2c32 Moved to doxyfile directory. 3f52204 Added doxygen configuration. ```git log --oneline v5.0.0-beta.1.. sources
``` 14866d8 fix: possible memory error in calling macros with arguments (#434) a5beea4 docs: remove invalid licence comment in extcmd.c (#431) 3188fbd fix: moebius_ function (#422, #430) 8165e80 fixing a few typos in sources folder 658bb45 fixing another bunch of typos in sources/ 7440741 Fix issue with implicitly or autodeclared dimension of autodeclared index cf39bb0 Allow setting ContinuationLines to 0, meaning infinity e5a9db9 Remove more gcc warnings (gcc version dependent) ad7be37 Put in the fix and undid another 3a6b7ec Undoes some of the previous commit to test the failing checks. e3f63b5 Repaired an inefficiency when many expressions are defined (TYPEEXPRESSION) b021935 Repairs the 'famous' GZIP bug. 8917fae Repair of the infamous gzip memory leak. ba2f606 Fixes #219 09b1f39 And repaired the problem with the check in the fix of #55 60af3fd Fixes #55 PrepPoly did not convert some arguments to fast notation. 56295bd Fix for issue #186 b7ba717 And yet another bug in TakeContent 91c8499 Fixed gcd_. Was problem with constant terms. f479692 Repaired a sortbot problem with big terms f5a216c Bucket fix and temporary collect % 415c646 Wrong variable was set to -1 on compiler error 6e8bd2f removed debug output f6973bd small fix in polynomial multiplication 7fd8d58 Changed the PUTNUMBER macro's to avoid warnings on 32-bits systems. 1915da7 fixed small memory leak in poly_to_argument 39b2ebd bugfix: if already factorized, don't factorize again 8661eeb ParFORM: fixed a bug of #write dbecf7f ParFORM: blocked statistics report from slaves c8776ed Little bugfix in %X option in #write. Counting problem. ```