[gentoo-dev] Re: proposal for consistency between {RUBY,PYTHON,PHP}_TARGETS

[Duncan]

Peter Stuge posted on Sat, 24 Nov 2012 20:20:27 +0100 as excerpted:

> Jauhien Piatlicki wrote:
>>> PHP_TARGETS="5.3 5.4"
>>> RUBY_TARGETS="1.9"
>>> PYTHON_TARGETS="2.7"
>>>
>>> But maybe it would be too problematic?
>>
>> What will you do with PYTHON_TARGETS="python2_7 python3_2 pypy1_9
>> jython2_5" then?
>
> That's an excellent point. Thanks!
>
> Thinking out loud another round: _TARGETS is an interface by Gentoo,
> so maybe it would not be such a bad idea to use existing Gentoo
> identifiers there, ie. (a subset of?) PMS version specifications.

On the net-nntp/pan upstream (which I've been involved with for about a 
decade now), but I'm sure it's not original to pan, wishlist bugs that 
would be nice to fix "someday", maybe when all the other bugs are fixed, 
or if someone profiles all the patches, does a bunch of testing, etc... 
these sorts of wishlist bugs are set to milestone target "bluesky".

IMO, that's exactly what this is, a "target bluesky" wishlist item.  
Except here it's worse, because the change will be very end-user visible, 
requiring configuration adjustments on running/working systems, for 
little reason, and unlike someone providing patches, someone can't 
reasonably volunteer to go around and fix everyone's systems for them.

Yes, it'd be nice to have consistent *_TARGETS values.  But IMO it's a 
whole lot of potentially bug triggering work on packages that are working 
just fine as they are, for comparatively little gain.  What's worse is 
that these changes will require end-user configuration changes.  So 
people aren't impressed with the inconsistency.  They'll be far LESS 
impressed if things break due to bugs, and I know a lot of former 
gentooers who already complain about both that, and about the need for 
constant attention to config changes, reading news and the various elog 
style notifications and jumping thru the necessary hoops to keep things 
working, etc.  We don't need MORE of those hoops to jump thru, and at 
this point, I just don't see that it's worth it.  Rather, it's almost at 
the level of change for change' sake, or at least, it's sure going to 
look like that to the users having to adjust their *_TARGETS vars.  
That's far less impressive than a bit of inter-package *_TARGETS 
inconsistency.

So like someone suggested in an earlier thread on simply changing some 
name or other, I take it if we're discussing this, all the REAL bugs are 
already fixed and there's nothing else more important left to do, right?  
Because that's about the point at which I think we should be focusing on 
things like this.

Just MHO, no more, no less.

-- 
Duncan - List replies preferred.   No HTML msgs.
"Every nonfree program has a lord, a master --
and if you use the program, he is your master."  Richard Stallman