Re: Why was handling of program_prefix changed?

[Akim Demaille]

>>>>> "Ian" == Ian Lance Taylor <address@hidden> writes:

Ian> Akim Demaille <address@hidden> writes:
>> :( It's bad that it only happens now.  Back then, the idea was
>> indeed to update everybody, and to simplify all the interfaces.
>> It's a pity the clarification of the semantics does not go up to
>> the top level :(

Ian> The semantics of the Cygnus configure script can be adjusted to
Ian> follow the new autoconf semantics.  However, the Cygnus configure
Ian> script invokes sub-configure scripts in different modes depending
Ian> upon how the subdirectory should be build--for the build machine,
Ian> the host machine, or the target machine.  The algorithm for doing
Ian> that used to be simple, and now becomes slightly more complex,
Ian> because we now must pass different sets of options to different
Ian> subdirectories.

I understand what it implies, and I repeat I'm sorry about it.

I don't feel like this ought to change, as it was decided after long
threads with people working in the area, and I'm definitely not one of
them.

My very humble opinion (as this is not a field I'm experienced in), is
that this agreement was very homogeneous with the whole set of
simplifications, in particular with the re-definition of
cross-compilation, and the re-definition of the chains of defaults
between build, host, target (which is precisely, target defaults to
host which defaults to build which defaults to config.guess).

At that time, it was clear someone will update the Cygnus configure,
and move that change up to the top level (I guess `top level' meaning
the user, just like for Autoconf configures).  I'm shocked to learn it
is still not done, and might have even been forgotten.  The most
alarming being the new chains of defaults, of course.

The latter will not change, as it faithfully reflects the GCS now, and
that was an unanimous change.

But wrt prefix_program, if *you*, Ian, tell me this change is wrong
(as opposed to good but implies more work elsewhere), then I'm revert
that bit.

But please, answer with what the initial specifications ought to have
been from the inception, not with history in mind.