❦ 24 décembre 2017 13:24 GMT, Chris Lamb <la...@debian.org> : >> Unrelated, but I am developing some kind of "lintian fatigue". […] >> Sometimes Lintian is right, sometimes it's not. > > As you imply, static analysis tools need to maintain a healthy signal- > to-noise ratio for them to remain relevant and useful. > > Needless to say, if Lintian is generating false positives for you, > please file bugs rather than ignoring all of its output! At the very > least, If one is seeing a problem, it is likely others are too.
I already often open or reply to bugs in lintian (including when I think severity is wrong). The main problem is not when lintian is wrong, the main problem if when lintian is right but is nit-picking. While I understand some of us would like to reach perfection, it is tiresome to fix every small issue, notably when they don't have any other effect than making a tool happy (and a few people) happy. And I never run lintian at pedantic level. I think we may loose contributors by trying to be perfect. As an example, the spelling errors are useful for debian/ directory (as informational), but totally useless for upstream stuff. For me, they are not worth telling upstream, they are not worth adding to an override (which could become outdated and give you another lintian warning). I have just updated a team-maintained package and I get: W: python-pyasn1: spelling-error-in-description-synopsis Python Python (duplicate word) Python W: python3-pyasn1: spelling-error-in-description-synopsis Python Python (duplicate word) Python Description: ASN.1 library for Python (Python 2 module) Description: ASN.1 library for Python (Python 3 module) Is that a bug? Dunno. I suppose someone with good intentions added that and now, I have either to open a bug report for lintian, add an override, fix the duplication or just ignore that hoping it will eventually go away. I see a discussion already happened about that: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=822504 https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=844166 Being perfect was a fine goal when we had 10k packages. We have too many of them now and we _have_ to maintain them because there are many dependencies to package for a single package of interest. We could afford to be perfect again in the future when we will accept anybody can fix those kind of problems without bothering the maintainer. People thinking this is important to fix those kind of problems can just do it themselves. Currently, the situation is that a few people can push their "agenda" (replace by a weaker word) to many people by pushing more stuff into Lintian (or in discussions in d-project or d-devel to "improve" packaging). And I know that you are open to both sides (I was able to make you revert a change, don't remember exactly which). -- The smallest worm will turn being trodden on. -- William Shakespeare, "Henry VI"
signature.asc
Description: PGP signature
