Bug#916735: lintian: appstream-metadata-missing-modalias-provide should be info, not warn
On Thursday, December 20, 2018 01:08:54 PM Petter Reinholdtsen wrote: > [Paul Wise] > > > I don't think I have the requisite time and understanding to do this, > > hopefully Petter will be interested to work on this but in general I > > think it will be best for individual upstreams to work on this since > > they know their software best and how to best expose which info. > > I registered a bts report against nitrokey-app with a proposed patch to > announce hardware support using appstream. See > https://bugs.debian.org/916911 >. > > I guess the documentation should be updated to make it more clear how to > do it, but I am a bit blind to exactly what is hard to comprehend about > the task, so I will need help with that. > > I do not have much spare time either, but did file quite a few bug > reports with patches for providing metadata on supported hardware to > several packages. Unfortunately most of these patches are just > lingering in BTS, so I ran out of steem and motivation for that > approach. Perhaps adding appstream information into policy is a better > approach. I kind of doubt it, as policy should should not be a stick to > force people to do work they do not want to do. How can me make people > want to provide hardware metadata for their packages? A lintian warning > help a bit, but it obviously not convincing the remaining package > maintainers. I appreciate you filing the bug. One more example of how I think the priority for this is wrong. Currently lintian claims this issue (W) is more important than missing hardening flags (I). I don't think that's right. I'll lay off this bug now as I think I've expressed myself fully and it's really up to the lintian maintainers to decided. Scott K
Bug#916735: lintian: appstream-metadata-missing-modalias-provide should be info, not warn
[Paul Wise] > I don't think I have the requisite time and understanding to do this, > hopefully Petter will be interested to work on this but in general I > think it will be best for individual upstreams to work on this since > they know their software best and how to best expose which info. I registered a bts report against nitrokey-app with a proposed patch to announce hardware support using appstream. See https://bugs.debian.org/916911 >. I guess the documentation should be updated to make it more clear how to do it, but I am a bit blind to exactly what is hard to comprehend about the task, so I will need help with that. I do not have much spare time either, but did file quite a few bug reports with patches for providing metadata on supported hardware to several packages. Unfortunately most of these patches are just lingering in BTS, so I ran out of steem and motivation for that approach. Perhaps adding appstream information into policy is a better approach. I kind of doubt it, as policy should should not be a stick to force people to do work they do not want to do. How can me make people want to provide hardware metadata for their packages? A lintian warning help a bit, but it obviously not convincing the remaining package maintainers. -- Happy hacking Petter Reinholdtsen
Bug#916735: lintian: appstream-metadata-missing-modalias-provide should be info, not warn
On Wed, 2018-12-19 at 07:28 +, Scott Kitterman wrote: > I'm not arguing it's a bad idea to have the check, but personally, I > get tired of looking at it. If this is important, get it in Policy > as a should and then I think warning would be appropriate. > > Why don't I just fix it? I read the referenced material on what > needs doing and concluded I don't have the spare mental cycles to > learn all about this for one package. In general I think it is perfectly acceptable to ignore lintian and other QA tools when one does not have the time or energy to make the changes that they are suggesting. I also think it is reasonable to override lintian for something you don't have time or energy for and don't want to see the suggestion any more. In the this case, I think that users just installing libnitrokey-common won't get anything useful, so a lintian override is appropriate here since it cannot know that a binary package (nitrokey-app) from a different source package (nitrokey-app) is the place to add the modalias metadata. nitrokey-app already has an AppStream file, but it doesn't have the modalias metadata. So I think that libnitrokey-dev needs to expose modalias metadata for nitrokey-app to export. > It'd be much more efficient for someone who both understands what > needs doing and cares to run through the affected packages and submit > patches. I don't think I have the requisite time and understanding to do this, hopefully Petter will be interested to work on this but in general I think it will be best for individual upstreams to work on this since they know their software best and how to best expose which info. -- bye, pabs https://wiki.debian.org/PaulWise signature.asc Description: This is a digitally signed message part
Bug#916735: lintian: appstream-metadata-missing-modalias-provide should be info, not warn
On Wed, 2018-12-19 at 10:57 +0100, Chris Lamb wrote: > Hi Paul, > > > Downgrading it to info level means that almost no-one will know about > > it, so you might as well just delete the tag instead. > > I don't agree with this view of "I" tags but playing severity wars is > not my idea of a good time so I've reverted this and I'll you folks > fight it out between yourselves.. I'm sorry for the tone in this part of my response, it was inappropriate. -- bye, pabs https://wiki.debian.org/PaulWise signature.asc Description: This is a digitally signed message part
Bug#916735: lintian: appstream-metadata-missing-modalias-provide should be info, not warn
tags 916735 - pending tags 916735 + moreinfo thanks Hi Paul, > Downgrading it to info level means that almost no-one will know about > it, so you might as well just delete the tag instead. I don't agree with this view of "I" tags but playing severity wars is not my idea of a good time so I've reverted this and I'll you folks fight it out between yourselves.. Regards, -- ,''`. : :' : Chris Lamb `. `'` la...@debian.org / chris-lamb.co.uk `-
Processed: Re: Bug#916735: lintian: appstream-metadata-missing-modalias-provide should be info, not warn
Processing commands for cont...@bugs.debian.org: > tags 916735 - pending Bug #916735 [lintian] lintian: appstream-metadata-missing-modalias-provide should be info, not warn Removed tag(s) pending. > tags 916735 + moreinfo Bug #916735 [lintian] lintian: appstream-metadata-missing-modalias-provide should be info, not warn Added tag(s) moreinfo. > thanks Stopping processing here. Please contact me if you need assistance. -- 916735: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=916735 Debian Bug Tracking System Contact ow...@bugs.debian.org with problems
Bug#916735: lintian: appstream-metadata-missing-modalias-provide should be info, not warn
On Wed, 19 Dec 2018 01:00:43 +0100 Chris Lamb wrote: > Nobody is doubting the value here, one just has to square that with > the idea that Lintian being too pedantic, noisy or making the wrong > priority choices is bad for effectiveness of tool in its entirity. :) There are only 50 packages affected by this tag, is it really that big of a problem for it to be at warning level? https://lintian.debian.org/tags/appstream-metadata-missing-modalias-provide.html Downgrading it to info level means that almost no-one will know about it, so you might as well just delete the tag instead. Looking at the package list there are only a few packages where the tag might not apply (like zfsutils-linux) and some of the overrides are definitely bogus. The package that Scott maintains that is affected by this tag (libnitrokey-common) contains udev rules for a specific set of USB devices, so it or another package (like nitrokey-app) definitely needs to have the modalias metadata declared so that users can easily find software for the Nitrokey and other devices when they plug them in. -- bye, pabs https://wiki.debian.org/PaulWise signature.asc Description: This is a digitally signed message part
Bug#916735: lintian: appstream-metadata-missing-modalias-provide should be info, not warn
On December 19, 2018 7:18:42 AM UTC, Paul Wise wrote: >On Wed, 19 Dec 2018 01:00:43 +0100 Chris Lamb wrote: > >> Nobody is doubting the value here, one just has to square that with >> the idea that Lintian being too pedantic, noisy or making the wrong >> priority choices is bad for effectiveness of tool in its entirity. :) > >There are only 50 packages affected by this tag, is it really that big >of a problem for it to be at warning level? > >https://lintian.debian.org/tags/appstream-metadata-missing-modalias-provide.html > >Downgrading it to info level means that almost no-one will know about >it, so you might as well just delete the tag instead. > >Looking at the package list there are only a few packages where the tag >might not apply (like zfsutils-linux) and some of the overrides are >definitely bogus. > >The package that Scott maintains that is affected by this tag >(libnitrokey-common) contains udev rules for a specific set of USB >devices, so it or another package (like nitrokey-app) definitely needs >to have the modalias metadata declared so that users can easily find >software for the Nitrokey and other devices when they plug them in. I'm not arguing it's a bad idea to have the check, but personally, I get tired of looking at it. If this is important, get it in Policy as a should and then I think warning would be appropriate. Why don't I just fix it? I read the referenced material on what needs doing and concluded I don't have the spare mental cycles to learn all about this for one package. It'd be much more efficient for someone who both understands what needs doing and cares to run through the affected packages and submit patches. In the meantime, I think info is the right level. Scott K
Bug#916735: lintian: appstream-metadata-missing-modalias-provide should be info, not warn
Petter Reinholdtsen wrote: > For this to work well, the appstream modalias data set need to be up > to date, complete and correct. Nobody is doubting the value here, one just has to square that with the idea that Lintian being too pedantic, noisy or making the wrong priority choices is bad for effectiveness of tool in its entirity. :) Best wishes, -- ,''`. : :' : Chris Lamb `. `'` la...@debian.org / chris-lamb.co.uk `-
Bug#916735: lintian: appstream-metadata-missing-modalias-provide should be info, not warn
[ Scott Kitterman] > The appstream-metadata-missing-modalias-provide relates to a nice to have > feature. Lintian treating this at a "Warning" level seems excessive for > something that's completely optional. Info seems right. The modalias appstream metadata provide a mechanism for users to locate the packages they need or want for their hardware. It can be compared to the package description used by 'apt search' and the file lists handled by 'apt-file search'. For this to work well, the appstream modalias data set need to be up to date, complete and correct. This make me conclude that 'warning' is the correct level for the lintian check. -- Happy hacking Petter Reinholdtsen
Bug#916735: lintian: appstream-metadata-missing-modalias-provide should be info, not warn
Package: lintian Version: 2.5.117 Severity: normal The appstream-metadata-missing-modalias-provide relates to a nice to have feature. Lintian treating this at a "Warning" level seems excessive for something that's completely optional. Info seems right. Scott K