Re: Severity of spelling errors
Russ Allbery wrote:
> Raphael Geissert writes:
>> Russ Allbery wrote:
>
>>> However, spelling errors are singled out as an example of a bug that
>>> should be minor severity by reportbug. Shouldn't we change those bugs
>>> to be severity: minor since we're matching Lintian tag severities with
>>> bug reporting severities?
>
>> I just re-checked the severity/certainty->EWI code translation table and
>> noticed that minor/certain tags are still represented by a warning, so
>> yes, I think the severity should be updated to reflect reality.
>
> They're currently all certain except spelling-error-in-binary (which is a
> bit of a special case since we may not be checking English text) and:
>
> copyright-file.desc:Tag: spelling-error-in-copyright
> copyright-file.desc-Severity: normal
> copyright-file.desc-Certainty: possible
> --
> manpages.desc:Tag: manpage-has-spelling-error
> manpages.desc-Severity: normal
> manpages.desc-Certainty: possible
> --
> menus.desc:Tag: spelling-error-in-doc-base-title-field
> menus.desc-Severity: normal
> menus.desc-Certainty: possible
> --
> menus.desc:Tag: spelling-error-in-doc-base-abstract-field
> menus.desc-Severity: normal
> menus.desc-Certainty: possible
>
> We seem to have a fairly good track record with the spelling checks, but
> the manpage one is brand new. I think the right thing to do is update the
> certainty to certain for all of them but the copyright check and the
> manpage check, with the intention of doing so with the manpage check as
> well after we have some confirmation that it's as reliable as the others.
Okay. I ran a quick test on multiple installed packages and did not find any
false positive, but until it is ran on the whole archive I'm not sure there
will be none.
>
> Looking at lintian.d.o, the copyright check does have a collection of
> false positives. als is a word in at least one non-English language used
> in some copyright documentation, and there are several overrides for
> misspellings used in licenses that we're reproducing verbatim and in
> e-mail communications with upstream. I'm therefore inclined to leave that
> one as possible and let it become an I: tag.
Okay.
>
> There's one other override, for spelling-error-in-doc-base-abstract-field,
> where the abstract is written in German where "programm" is a word.
This might actually be treated as a bug in doc-base for not providing a way
to differentiate between documentation in different languages.
>
> The only other one that I'm unsure about is the description test, where a
> certainty of certain is a bit of a reach given the number of legitimate
> overrides there are (usually about Lintian's capitalization checks wanting
> to capitalize program names).
>
Not sure if this check should be split to better reflect the certainty of
both checks.
>
>> P.S. have you had time to check some of the infrastructure changes I
>> proposed?
>
> Not yet, but I'm very close. I've been spending the vacation digging
> myself out of an e-mail hole. :)
>
Heh, okay :)
Cheers,
--
Raphael Geissert - Debian Developer
www.debian.org - get.debian.net
Reply to: