Here is a patch that allows projects using sources derived using
suffix rules (corba, esql, whatever...) to be built in parallel.
The initial discussion which lead to this patch is available from the
list archives ( try
http://sourceware.cygnus.com/ml/automake/2000-03/msg00112.html ). I
basical
On Mar 29, 2000, Tom Tromey <[EMAIL PROTECTED]> wrote:
> I'm skeptical that this approach will work :-(
I still think we should revert to the old behavior of Makefile
depending on BUILT_SOURCES, at the very least, as a deprecated
feature, for backward compatibility. Otherwise, builds may simply
On Mar 29, 2000, Tom Tromey <[EMAIL PROTECTED]> wrote:
>>> I'm skeptical that this approach will work :-(
Alexandre> I still think we should revert to the old behavior of
Alexandre> Makefile depending on BUILT_SOURCES, at the very least, as
Alexandre> a deprecated feature, for backward compatibil
Alexandre> So maybe we should use a stamp file that all $(OBJECTS)
Alexandre> depend upon, and that runs $(MAKE) $(BUILT_SOURCES) before
Alexandre> it touches itself. It still wouldn't work in case someone
Alexandre> removes one of the BUILT_SOURCES but leaves the stamp file
Alexandre> around, bu
>> I'm skeptical that this approach will work :-(
Alexandre> I still think we should revert to the old behavior of
Alexandre> Makefile depending on BUILT_SOURCES, at the very least, as
Alexandre> a deprecated feature, for backward compatibility.
This only works for GNU make. So it was never a r
> "Alex" == Alex Hornby <[EMAIL PROTECTED]> writes:
Alex> Do you have any hints before I dive in? Is the current CVS head
Alex> a good starting point?
CVS head is a good starting point.
There is a HACKING file that explains some rules (which may or may not
apply; I forget exactly what is in
Alexandre/Tom,
If no one else is working on it I'm interested in implementing this
hook functionality. Parallel builds will save me a significant amount
of time.
Do you have any hints before I dive in? Is the current CVS head a good
starting point?
TIA
Alex.
Alexandre Oliva writes:
> On Mar
On Mar 22, 2000, Tom Tromey <[EMAIL PROTECTED]> wrote:
> This might be the best way to go.
> It is sort of a pain to implement; for each such exe we'd have to do
> this:
> exe: dummy
>build the exe
How about
$(exe_OBJECTS): exe-hook
--
Alexandre OlivaEnjoy Guaraná, see http:/
> "Alex" == Alex Hornby <[EMAIL PROTECTED]> writes:
>> I agree it would be nice to say "build these files before doing
>> anything else". That's sort of a pain with make. One idea would
>> be to introduce an "all-hook" which would be run before "all".
>> This solves the common problem, but
Tom Tromey writes:
> > "Alex" == Alex Hornby <[EMAIL PROTECTED]> writes:
>
> >> How about just creating standard Makefile dependencies?
> >>
> >> Foo_impl.o: Foo_s.hh Foo_c.hh Foo_s.cpp Foo_c.cpp
> >> Foo_impl.$(OBJEXT): Foo_s.hh Foo_c.hh Foo_s.cpp Foo_c.cpp
>
> Alex> That would wor
> "Alex" == Alex Hornby <[EMAIL PROTECTED]> writes:
>> How about just creating standard Makefile dependencies?
>>
>> Foo_impl.o: Foo_s.hh Foo_c.hh Foo_s.cpp Foo_c.cpp
>> Foo_impl.$(OBJEXT): Foo_s.hh Foo_c.hh Foo_s.cpp Foo_c.cpp
Alex> That would work. I wanted to avoid adding dependencies ma
Alexandre Oliva writes:
> On Mar 21, 2000, Alex Hornby <[EMAIL PROTECTED]> wrote:
>
> > I would like to ensure that Foo_[sc].hh and Foo_[sc].cpp are present
> > before Foo_impl.cpp is built for the first time.
>
> How about just creating standard Makefile dependencies?
>
> Foo_impl.o: F
On Mar 21, 2000, Alex Hornby <[EMAIL PROTECTED]> wrote:
> I would like to ensure that Foo_[sc].hh and Foo_[sc].cpp are present
> before Foo_impl.cpp is built for the first time.
How about just creating standard Makefile dependencies?
Foo_impl.o: Foo_s.hh Foo_c.hh Foo_s.cpp Foo_c.cpp
Foo_impl.$(
> "Alex" == Alex Hornby <[EMAIL PROTECTED]> writes:
Alex> This is a problem as Foo_real.cpp includes Foo_derived.h.
Alex> How can I get a result similar to the old BUILT_SOURCES, where
Alex> the commands to produce the derived source files for a target
Alex> are run _before_ the non derived
When I perform a parallel (MAKE="gmake -j4") build for derived
sources, the derived sources are not always produced before the non
derived sources are built. This is a problem as Foo_real.cpp includes
Foo_derived.h.
Perhaps this is a make problem? I'm using gnu make 3.76.1 and automake
1.4.
Ho
15 matches
Mail list logo
