On Wed, Apr 21, 2021 at 11:02:42PM +0530, Pirate Praveen wrote: > >This is not the spirit of HEAD/pretty mode: the goal is to pick commits > >after last tag. Upstream did a strange thing in this repo: set a tag > >outside any named branch. I'm not sure we should modify uscan because of > >an unlikely upstream behavior.
I do agree this is a very odd behaviour, and likely a unique one. In
this case it's not a forgotten tag, is a tag that is not an ancestor of
what you downloaded. (that said, I can't see such tag in my clone of
bnjmnt4n/lodash-cli ? So, what are you talking about?)
> >But if Devscript Team agree with you, I can modify "ctype" feature to
> >fix tag when last tag is lower than package.json#version, then version
> >will be 4.17.5+timestamp instead of 4.17.4.20+timestamp
>
> I have reopened for comments from devscripts team.
So my "official" on this is: before getting yet another feature like
this, I'd like at least another package where this actually proves
useful and preferably more than one more.
> It does not have to be the default, but as an optional setting in the watch
> file. May be ctype=nodejs,version=package.json
>
> Or ctype=nodejs,pretty=package.json
>
> In this case it should be 4.17.21+git.timestamp.hash as version in
> package.json is 4.17.21.
Besides, in this case it's not even that useful, IMHO.
Since you are using the group+checksum feature, the actual version that
uscan gets from the lodash-cli component is very hidden and doesn't
really matter in the end, as long it monotonically increases whenever
there is an updated upstream, which I believe it does in your watchfile.
> uscan supports a lot of weird upstream conventions anyway. Many upstream
> don't use tags consistently so we need ways to handle those cases.
Well, I would rather we do not add more support for more weird
upstreams. There used to be a time when we tried to collaborate with
upstreams to get something that works for both, and I honestly believe
that asking them to create tags whenever they release something is
totally acceptable and you should do that. Did you try here? What
answer did you get?
> Also checksum option does not support mode=git scheme default values. So I
> used pretty=4.17.21.%cd to force using digits only.
>
> Should I open another bug for using checksum with git ? Current it supports
> only digits in version so ~git or +git or the hash in version does not work.
Well, for this, whatever Yadd prefers. I guess forcing a different
pretty= format with version `checksum` is fine to do and probably makes
sense if it fails otherwise.
--
regards,
Mattia Rizzolo
GPG Key: 66AE 2B4A FCCF 3F52 DA18 4D18 4B04 3FCD B944 4540 .''`.
More about me: https://mapreri.org : :' :
Launchpad user: https://launchpad.net/~mapreri `. `'`
Debian QA page: https://qa.debian.org/developer.php?login=mattia `-
signature.asc
Description: PGP signature
