tech-pkg archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Re: CVS commit: pkgsrc/net/gssdp

John Marino <> writes:

> On 12/4/2016 16:13, Joerg Sonnenberger wrote:
>> On Sun, Dec 04, 2016 at 04:12:03PM -0600, John Marino wrote:
>>> On 12/4/2016 16:04, Joerg Sonnenberger wrote:
>>>> On Sun, Dec 04, 2016 at 09:56:30PM +0000, John Marino wrote:
>>>>> Module Name:	pkgsrc
>>>>> Committed By:	marino
>>>>> Date:		Sun Dec  4 21:56:30 UTC 2016
>>>>> Modified Files:
>>>>> 	pkgsrc/net/gssdp: Makefile
>>>>> Log Message:
>>>>> net/gssdp: Specify readline and ncurses requirements
>>>>> Solves:
>>>>> /usr/libexec/binutils225/elf/ error: cannot find -lreadline
>>>>> /usr/libexec/binutils225/elf/ error: cannot find -lncurses
>>>> Please revert this and the earlier changes. Nothing in this package or
>>>> e.g. security/pam-krb5 uses readline directly. At most it gets pulled in
>>>> directly from somewhere and should be fixed there and only there.
>>> # ldd ./
>>> ./
>> Yes, please read again what I wrote. One of the dependencies pulls in
>> readline, but doesn't for whatever reason have it in its
>> That is the problem that should be fixed.
> So you can fix that.  Once you've identified exactly what it is, then
> tell me and I'll verify those recently changed ports still build when
> readlink buildlink is removed.  It makes no sense to intentionally
> break ports until then.  Being linked to products == used DIRECTLY
> though.

I'm having a bit of trouble following.  Joerg's point that only packages
that actually link against readline should have a bl3 line seems
correct.  And, for packages that don't intend to have readline, if they
do link against it without it being present, that seems like a bug that
should be fixed (perhaps with --disable-readline).   Also, a change of
this magnitude seems like it should be proposed/discussed; at least in
this case there are differing technical opinions.

Are you saying that you have added readline as a bl3 dependency to
packages where you have not verified that the package actually directly
links against readline?

(not speaking as pmc, and this is not yet a request.)  Overall, it seems
like it might be better to revert and to see if the group can find a
solution with less collateral damage.

Home | Main Index | Thread Index | Old Index