On Mon, Sep 07, 2020 at 06:56:08PM -0700, Felix Lechner wrote: > > lintian-brush can use this as a signal to fix typos in field names > > I'll try to track down a list of valid names and will happily implement it. There's a list in the specification at https://wiki.debian.org/UpstreamMetadata
> In a pinch, you could even do it yourself. The tag > upstream-metadata-field-present [1], which as a classification tag is > not shown for packages on the website but is available in UDD, gives > you the verbatim field names found. Like me, you just need a list of > valid names to determine those that are bogus. They are already only > one SQL query away! The classifications are super useful, and having them in UDD is especially great for finding information about packages across the archive, e.g. "which packages use a Vcs-Browse URL that's not on HTTP/HTTPS?" There is a lot of value in keeping the business logic for finding problems (in this case, to determine invalid field names) in lintian rather than just having it bubble up raw data, i.e. as the arbiter of what is right or wrong: * the business logic for finding a problem is only in a single place * lintian-brush is opportunistic and only goes after issues that it can fix. it's valuable that humans get told about all problems so they can fix those things that need manual operation * it allows lintian to be used to verify that an issue has been fixed after lintian-brush has been run Cheers, Jelmer -- Jelmer Vernooij <[email protected]> PGP Key: https://www.jelmer.uk/D729A457.asc
