libtool-patches
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: ideas for speeding up compile mode


From: Ralf Wildenhues
Subject: Re: ideas for speeding up compile mode
Date: Sun, 15 May 2005 08:37:22 +0200
User-agent: Mutt/1.5.9i

Hi Ralf,

* Ralf Corsepius wrote on Sun, May 15, 2005 at 07:44:40AM CEST:
> On Fri, 2005-05-13 at 17:06 +0200, Ralf Wildenhues wrote:
> > [ CC to automake because it is involved -- see below ]
> 
> > 3) Fold compile mode completely into Automake snippets/the resulting
> > Makefile.  I wonder why this has not been done before, at least for the
> > default rules (per-target rules could lead to a large blowup of the
> > resulting Makefile.in).
> Such kind of proposal had popped up several before.

That's what I thought, but not found.

> IIRC, the argument had been that the libtool folks had wanted to "keep
> libtool outside of automake" because libtool is/had been supposed to be
> a tool independent of the make infrastructure being used (automake is
> just one of them).

OK.  Well, I've a couple of arguments for it which I'd like weighed
against the other solutions I suggested:

First, as I already said, we could provide support for other make
infrastructures as well.  Second, the proposal did not include removing
compile mode from `libtool', so nobody really loses.  Third, it may
enable us to actually profit from the fact that we don't need to compile
both PIC and non-PIC in some cases.

If you ask me, the increased number of special cases does not really add
much to the overall complexity.  Libtool already has an abundance of
special cases, many of which are more difficult to check.

The bug report by the libjava people suggests a possible speedup of
almost 2/3 resp. 5/6 for their case.  To me, this proposal "just sounds
like the right thing", but I am certainly open to arguments.

Regards,
Ralf




reply via email to

[Prev in Thread] Current Thread [Next in Thread]