[MacPorts] #50204: There is no need for subports, they are actually harmful

[MacPorts]

#50204: There is no need for subports, they are actually harmful
--------------------------+-------------------------
  Reporter:  nshmyrev@…   |      Owner:  michaelld@…
      Type:  enhancement  |     Status:  closed
  Priority:  Normal       |  Milestone:
 Component:  ports        |    Version:
Resolution:  wontfix      |   Keywords:
      Port:  swig         |
--------------------------+-------------------------
Changes (by michaelld@…):

 * status:  new => closed
 * resolution:   => wontfix


Comment:

 I don't know the history of how or why the SWIG ports were split and
 distributed in the manner they are. If I had to guess, it's because some
 other package managers do it this way & it's really convenient for precise
 dependency tracking (e.g., depending not just 'swig', but on the Python
 bindings from 'swig-python'). If we just installed "swig" with everything,
 there would be no way to actually have reasonable belief and/or robustly
 verify that the bindings were installed properly.

 Making a separate port increases the chances that the bindings were
 properly installed -- if for some reason the build fails or does not
 install anything, then 'port' will error out & it will be obvious that the
 bindings failed (for some reason). I like the specific dependency tracking
 allowed by keeping the language bindings separate from the primary port.

 Although some of the swig* ports might have no dependencies, or could
 possible use System dependencies, the "MacPorts way of doing things" does
 not mean that we should just be installing them anyway with SWIG.

 SWIG is pretty easy to maintain as it currently is because someone (not
 me) did a great job writing the Portfile in the first place. Given the
 amount of work that would be required to go back to a single swig install,
 I'm not inclined to go there without some -very- persuasive arguments to
 do so. I'm not convinced by your arguments here, especially the note using
 binding ports is harmful. Having binding ports is just another way of
 splitting installs, and it comes with some advantages over the
 alternative. Maybe I'm just not understanding your arguments?

 I'm going to close this ticket as "won't fix". My advice if you truly feel
 strongly about this matter is to bring up your issue on the MacPorts
 developer email discussion group & refer to this ticket. If you can
 convince the MPDevs that this change is appropriate, good, and necessary,
 then we'll find a way to make it happen.

 [As a related aside: A few years back, I maintained various gnuradio-*
 subports, which was a real pain when a major version change happened &
 some subports needed to be removed while others were added. I queried GR
 users and developers if they cared about the supports and the vast vast
 majority said they would prefer a single install; thus, I made it so &
 it's been -so- much easier for me to maintain since then. Thus, moving
 from multiple subports to a single port can be done, if one gets the
 correct info & uses it well.]

-- 
Ticket URL: <https://trac.macports.org/ticket/50204#comment:2>
MacPorts <https://www.macports.org/>
Ports system for OS X