On Sat, 2018-08-11 at 22:06 -0400, Robert Marcano wrote:
> On Sat, Aug 11, 2018 at 8:07 PM Rex Dieter <rdie...@math.unl.edu> wrote:
> > Robert Marcano wrote:
> > 
> > > For example, someone developing against krb5-devel for a GSSAPI client,
> > > probably doesn't need openssl-devel installed, that they are linking
> > > against Kerberos doesn't means they use the same crypto library
> > > directly, they could use nss for example.
> > 
> > I think you're most likely going to need to deal with this on a
> > case-by-case
> > basis.  In this specific example, find out for sure if "probably doesn't
> > need openssl-devel" is entirely accurate or not, and take appropriate
> > measures.
> > 
> Not sure a case-by-case basis will always work, for example I force removed
> compat-openssl10-devel (rpm - e compat-openssl10-devel --nodeps) that is
> pulled by nodejs-devel, and I am able to link against NodeJS libraries,
> because the modules I need to build doesn't use OpenSSL. So
> compat-openssl10-devel should not be a hard dependency for nodejs-devel.
> Sadly the package maintainer think this is fine [1].
> A small packaging guideline change about reducing *-devel hard dependencies
> when they aren't  always required could help.

You're treating complicated situations as if they were simple, which is
always a bad idea. Reality does not always map perfectly to "requires"
or "does not require". Stephen explained in the bug why the dependency
is present, and it's a good reason: if it's not there, Node modules
could well get built such that they don't work properly. You're not
making a very good case for why this dependency is wrong, let alone why
there should be a guideline saying it shouldn't exist.
Adam Williamson
Fedora QA Community Monkey
IRC: adamw | Twitter: AdamW_Fedora | identi.ca: adamwfedora
devel mailing list -- devel@lists.fedoraproject.org
To unsubscribe send an email to devel-le...@lists.fedoraproject.org
Fedora Code of Conduct: https://getfedora.org/code-of-conduct.html
List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
List Archives: 

Reply via email to