Re: libf2c FLAGS_TO_PASS [was Re: egcs-1.2 stuff ]

[Jeffrey A Law <law@upchuck.cygnus.com>]

  In message < 19990415135929.23354.qmail@deer >you write:
  > If I get to this, I'll try a build with the patch in, and maybe play
  > a little.  If I don't see a problem, why not just check it in, and
  > see if it breaks things?
Well, that's not really the attitude we want to have.  Though I do think I
understand what you're trying to say :-)

  > I take Dave's concerns seriously, but if I
  > can't get a clear description of the problems double-quoting would
  > introduce, sufficient to override Jeff's valid point about consistency
  > wrt other libraries, then that might be the time to bazaarize the
  > patch and see whether a bunch more eyes help.
It looks to me the problem is consistency when we run config.status after
something like configure changes.

What I don't get is why we pass these things around when doing that anyway,
isn't this stuff actually embedded in config.status?  Is it right for a
rebuild to use potentially different values for those variables?

Or is the setting of those variables solely to make the _tests_ run by
configure/config.status work properly?  In which case we probably can settle
on a set of variables that need to be passed with the different quoting
scheme and arrange to pass them to all the runtime libraries so that we don't
have to always go into the runtime lib dirs if we need to update that list.

Depending on the various answers and the amount of work each entails we may
take different directions -- for example, we might decide the issue simply
isn't important enough to risk revamping this stuff this close to egcs-1.2.
Or we may decide that HJ's patch is good enough for egcs-1.2.  Or we may
decide that we can unify this particular issue in the runtimes easily and do
it now.  Or just about anything in between.

The important thing is to get our facts & options together quickly so that we
can make informed decisions.


jeff