Re: Dependencies on www/zope3

[Krister Walfridsson <krister.walfridsson%gmail.com@localhost>]

On Wed, 21 Aug 2013, Greg Troxel wrote:

> Why not let them be?  Could you explain why you think it's in the best
> interest of the greater pkgsrc ecosystem to remove things from wip just
> because of a danglingd dependency, rather than leave them for someone to
> use as a base later?  (Certainly it seems fine to drop them from the
> top-level wip/Makefile.)

Well, I do not know -- that is why I asked "What is the right way to 
handle this?" I.e. doing nothing could be the right way...  :)

But I think we should be cleaning out broken, abandoned packages.  The 
reason is that
1. I do not think they are useful,
2. they come with a cost.


1. Usefulness
-------------
I do not think it in general is useful to use old, broken, abandoned 
pkgsrc-wip packages as a base for an updated package.  To begin with, I 
expect the old package to be work-in-progress, i.e. it is unclear if it 
ever worked.  And the "old, broken, abandoned" packages are old, which 
means that there usually have been big changes in both the pkgsrc 
infrastructure and in upstream source code, so you often should do things 
in a different way anyway...


2. Cost
-------
The pkgsrc-wip is used to get new people involved, and for a temporary 
area when creating packages to import to pkgsrc.  I think both of these 
usages could be helped by frequent bulk builds for a range of different 
targets, and my goal has been to set up such a build farm.  But all these 
abandoned packages makes this need much work than necessary.  The reason 
I'd like to fix/remove the zope3-using packages is that this failure 
prevents the bulk build from starting.  There are about 30 other packages 
that also blocks bulk building, and I'd like all of them fixed so that 
"anyone" can run the bulk builds without needing to manually exclude the 
problematic packages.  (And there are other annoyances such as the old 
wip/glib2 that prevents most of the packages to build, that also should be 
fixed...)

And changes in pkgsrc introduces problems for the abandoned packages that 
need to be investigated (and often blocks the build from starting).

The reason to do these bulk builds is to find issues in the actively 
worked on packages.  But these failures risk getting lost in the noise 
from the >1000 abandoned packages that fail.

And all these broken, abandoned packages drag in lots of packages from 
pkgsrc as dependencies, which means that it is much slower to build than 
necessary as the bulk builder need to build most of pkgsrc too (which 
means that I cannot run the bulk build on some of my slower machines, such 
as sparc64).

Most developers seems to try to update pkgsrc-wip too when doing 
infrastructure changes, moving packages from wip to pkgsrc, etc. I.e. it
need extra work that better could be spent on real improvements.

    /Krister

------------------------------------------------------------------------------
Introducing Performance Central, a new site from SourceForge and 
AppDynamics. Performance Central is your source for news, insights, 
analysis and resources for efficient Application Performance Management. 
Visit us today!
http://pubads.g.doubleclick.net/gampad/clk?id=48897511&iu=/4140/ostg.clktrk
_______________________________________________
pkgsrc-wip-discuss mailing list
pkgsrc-wip-discuss%lists.sourceforge.net@localhost
https://lists.sourceforge.net/lists/listinfo/pkgsrc-wip-discuss