Hi Joe, On 2022-09-14 09:09:25 -0700, Joe Perches wrote: > On Wed, 2022-09-14 at 12:02 +0200, Niklas Söderlund wrote: > > Add a warning for fixes tags that does not follow community conventions. > [] > > * Changes since v6 > > - Update first check to make sure that there is a likely SHA1 of some > > minimum length after the fixes line. > > https://lore.kernel.org/lkml/2febb7893346b6234983453de7c037536e479bfc.camel@xxxxxxxxxxx/ > > The goal here should be to identify a line that looks like a commit > reference. > > So find lines that starts with 'fixes' and have a SHA1 commit id as > broadly as reasonable. > > Did you run the grep pattern and look at the results? > > One grep pattern to verify the non canonical fixes format that > are mistakenly used is: > > $ git log --since=5-years-ago --no-merges --grep='^\s*fixes' -i --format=email -P | \ > grep -P -i '^\s*fixes' | \ > grep -P -v '^Fixes: [0-9a-f]{12,12}\s*\(".*")' > > [] > > There are many different styles. > Parenthesea are sometimes not used. I understand this, and I did have a look at it. > > > + if ($line =~ /(\s*fixes:?)\s+([0-9a-f]{5,})\s+($balanced_parens)/i) { > > How about some pattern like > > /fixes\s*:?\s*(?:commit:?\s*)?[0-9a-f]{5,}/i > > or maybe even more broadly: > > /fixes\b.*\b[0-9a-f]{5,}\b/i Maybe I misunderstand your comment, but this is what I do in this patch? if (!$in_header_lines && $line =~ /^\s*fixes:?\s*(?:commit\s*)?[0-9a-f]{5,}\b/i) { ... if ($line =~ /(\s*fixes:?)\s+([0-9a-f]{5,})\s+($balanced_parens)/i) { ... } } This will catch and warn about such tags but not attempt to break out it's components in order to suggest a potentially more correct fix. Is it this second filter you would like me to change? -- Kind Regards, Niklas Söderlund
