----- Original Message -----
From: "Radek Holy" <rholy(a)redhat.com>
To: "James Antill" <james(a)fedoraproject.org>
Cc: "Development discussions related to Fedora"
<devel(a)lists.fedoraproject.org>
Sent: Saturday, February 14, 2015 7:53:50 PM
Subject: Re: another dnf kernel issue?
----- Original Message -----
> From: "James Antill" <james(a)fedoraproject.org>
> To: "Radek Holy" <rholy(a)redhat.com>
> Cc: ndbecker2(a)gmail.com, "Development discussions related to Fedora"
> <devel(a)lists.fedoraproject.org>
> Sent: Friday, February 13, 2015 8:28:44 PM
> Subject: Re: another dnf kernel issue?
>
> On Tue, 2015-02-10 at 04:01 -0500, Radek Holy wrote:
>
> > TBH, I don't know whether we should extend the forms of package
> > specifications to support your case. The current behaviour seems to be
> > safer to me. I mean, if we improve it, user wouldn't be able to query
> > just package names as easily as now.
>
> Safer? I can't think how.
> FWIW, in yum we did it the other way and added install-n, remove-n for
> just operating on the names of packages. Seemed less confusing if you
> want to force it, and did what people expected. YMMV.
With the current syntax, you can limit the globs to package names and still
you can append version or architecture specifications (globs or not). So
there is lower probability that you select packages that you didn't want to
select (e.g. packages containing numbers in their name). That's why I
consider it safer.
So with our syntax you can more easily express yourself and so far I don't
know about anything that can be expressed via YUM's globs but not via DNF's.
And also we don't need yet another command.
Anyway, I'm not trying to defend the current DNF syntax. This is just my
opinion. And TBH I didn't think about it too much. I don't think this
discussion is very much needed.
Feel free to correct me and explain me why this issue is important. It's definitely
possible that I've missed something.
--
Radek Holý
Associate Software Engineer
Software Management Team
Red Hat Czech