Re: pkg/55912: gstreamer1 incomplete package installation / failed build

[nia <nia%NetBSD.org@localhost>]

The following reply was made to PR pkg/55912; it has been noted by GNATS.

From: nia <nia%NetBSD.org@localhost>
To: gnats-bugs%netbsd.org@localhost
Cc: 
Subject: Re: pkg/55912: gstreamer1 incomplete package installation / failed
 build
Date: Wed, 6 Jan 2021 12:49:20 +0000

 On Wed, Jan 06, 2021 at 11:10:01AM +0000, Aleksey Arens wrote:
 >  To me, as a system administrator, the presence of a *vanilla*
 >  `introspection' signifies that there's yet more bloat to be removed.  If
 >  those features turn out to be essential for some uses in important
 >  dependencies, then it makes sense for them to be treated as not really
 >  options, but rather as platform workarounds.  This is especially so, if
 >  they do not add heavy dependencies (they really don't), or questionable
 >  dependencies (like e.g. dbus -- and they don't), and don't affect the
 >  performance adversely (that I'm not absolutely 146% sure about, but I would
 >  assume since they're being merely type descriptions, and since they must be
 >  describing generic datastructures, the performance should not be affected).
 
 Disabling introspection might break dependent packages if they're written
 in scripting languages, AFAIK it's the way GObject stuff exposes bindings.
 
 The decision to angrily add introspection options everywhere was made
 some time ago, and maybe it's time to move on.
 
 >  Finally, there's a question, as to how gstreamer1 managed to be built on
 >  platforms with introspection disabled in its dependencies?  In those cases,
 >  the build must have failed, especially with
 >  PKG_DEFAULT_OPTIONS+=-introspection.
 
 Might not have been tested in a while and package updates happened since.