Re: GNU make 3.81beta4 released

[Markus Mauhart]

"Eli Zaretskii" <address@hidden> wrote ...
>
> > From: "Markus Mauhart" <address@hidden>
> > Date: Sat, 14 Jan 2006 15:22:23 +0100
> >
> > "Eli Zaretskii" <address@hidden> wrote ...
> > >
> > > It is IMHO wrong to have README.W32 mention the CVS build except
> > > with a single sentence which tells to look in README.cvs.
> >
> > This is exactly what I did.

I still think this is true, i.e. IMHO my README.W32 is not CVS skewed
at all, but in the end READMEs are a matter of taste.


> > Additionally you yourself felt necessary to warn me in a previous posting:
> >
> > > Make sure you delete config.h before you do this, so that
> > > build_w32.bat will copy config.h.W32 to config.h and use that.
>
> That's because you created config.h with a different procedure, and I
> told how to build it with the ``normal'' procedure.  In effect, that
> required to undo your configuration.  You could have deleted config.h
> instead of copying config.h.W32.
>
> > IMHO with my suggested changes in README.W32[.template] this warning is
> > no more necessary.
>
> It is not necessary at all to warn about it, when people follow the
> build procedure.

No: the existing build procedure talks about editing config.w32.h, but
build_w32.bat doesnt necessarily use it. Hence your warning is necessary
with the existing README.W32.


> > > How close? should I stop working on the -jN issue on Windows?
> >
> > No :-)
>
> You will see that I just posted the patches for that.


I have tried them out.
Unfortunately for the project I'm currently interested I get the same strange
behaviour as recently when I for myself removed the if-w32-disable-j-code.
It seems to be a combination of my "complicated" makefile stressing gmake
generally, then stressed w32-code paths, stressed "-j"-codepaths, ...
Without "-j [n] under w32", it workes flawless since 1 year, nevertheless
it may be some bug in my makefile.


1) inside cygwin, "make -f Mymakefile -j 2or3orNothing" works as I expect, i.e
the corresponding number of concurrent compiles/links happen.

2) inside cmd.exe, using  my trivial or your current patch:
"make -f Mymakefile -j 2or3orMore" ... executes like no "-j" had been used.
"make -f Mymakefile -j"
two kinds of nonsense happen:

Mymakefile running in mode 1, make creates 100-1000 bat files, executes
10-100 of them, then make.exe hanges, consuming its complete CPU, having no
process childs..

Mymakefile running in mode 2, make creates some 100 bat files, then ...

$ make.cvs.20060114.j_patch.mss.exe -f out\themake.mk cfG=vc pkG=ide 
upp.linkmode=1 upp.blitz=0 -j
make.cvs.20060114.j_patch.mss.exe[1]: Entering directory `out' touch
out/plugin/jpg/vc.Gui.Sh/jdmerge.obj touch out/plugin/jpg/vc.Gui.Sh/jdphuff.obj 
touch
out/plugin/jpg/vc.Gui.Sh/jdpostct.obj touch out/plugin/jpg/vc.Gui.Sh/jdpred.obj 
touch
out/plugin/jpg/vc.Gui.Sh/jdsample.obj touch 
out/plugin/jpg/vc.Gui.Sh/jdscale.obj touch
out/plugin/jpg/vc.Gui.Sh/jdshuff.obj touch out/plugin/jpg/vc.Gui.Sh/jdtrans.obj 
touch
out/plugin/jpg/vc.Gui.Sh/jerror.obj touch out/plugin/jpg/vc.Gui.Sh/jfdctflt.obj 
touch
out/plugin/jpg/vc.Gui.Sh/jfdctfst.obj touch 
out/plugin/jpg/vc.Gui.Sh/jfdctint.obj touch
out/plugin/jpg/vc.Gui.Sh/jidctflt.obj touch 
out/plugin/jpg/vc.Gui.Sh/jidctfst.obj touch
out/plugin/jpg/vc.Gui.Sh/jidctint.obj touch 
out/plugin/jpg/vc.Gui.Sh/jidctred.obj touch
out/plugin/jpg/vc.Gui.Sh/jmemansi.obj touch 
out/plugin/jpg/vc.Gui.Sh/jmemmgr.obj touch
out/plugin/jpg/vc.Gui.Sh/jpg_opt.obj touch out/plugin/jpg/vc.Gui.Sh/jpgupp.obj 
touch
out/plugin/jpg/vc.Gui.Sh/jquant1.obj touch out/plugin/jpg/vc.Gui.Sh/jquant2.obj 
touch
out/plugin/jpg/vc.Gui.Sh/jutils.obj touch 
out/plugin/ndisasm/vc.Gui.Sh/disasm.obj
make.cvs.20060114.j_patch.mss.exe[1]: touch: open: 
out/plugin/ndisasm/vc.Gui.Sh/disasm.obj: No such
file or directory make.cvs.20060114.j_patch.mss.exe: Interrupt/Exception caught 
(code = 0xc0000005,
addr = 0x398) make.cvs.20060114.j_patch.mss.exe: *** [all] Error 255

Maybe I'll do additional experiments this weekend ...


Best Regards
Markus.