Hi!

On Wed, 2026-01-28 at 00:42:22 +0100, Guillem Jover wrote:
> On Wed, 2026-01-28 at 00:30:21 +0530, Nilesh Patra wrote:
> > Package: dpkg
> > Severity: normal
> > X-Debbugs-Cc: [email protected]
> > Control: affects -1 + src:lintian
> > Version: 1.23.5
> 
> > While trying to fix tests for lintian, I noticed that specifying the
> > email as "username@localhost" in Uploaders field does not cause the
> > build to choke.
> > 
> > That means that dpkg is accepting malformed address here, as it should
> > check also if TLD is present or not.
> 
> While implementing the parser/validator I wondered whether I should
> enforce this, and because I was unsure, I left it as is (at least for
> now), as we can always enforce it later (given that there was no
> enforcing before).
> 
> I guess my worry would be whether that would break any existing valid
> use (outside of the Debian archive) for locally/internal/vendor built
> packages, and for the myriad of brand TLDs (for example), although for
> the latter I think I've read somewhere that TLDs are not supposed to
> have MXs defined?

I've been thinking about this, and ISTM that this has the potential
for lots of local breakage, in test suites, CI runs or whatever. So
I'm not comfortable adding a fatal check for this. I'm fine though
with a warning, so I'll do that instead.

If lintian expects this to be an error for example in the Debian
context, then I'd say it should keep a test for that.

Thanks,
Guillem

Reply via email to