Re: improvements for parallel makes ?

[Christophe LYON]

Boris Kolpackov wrote:
> Alexey,
>
> Alexey Neyman <address@hidden> writes:
>
> > Thanks for the interest in this feature. As you might have seen from
> > further discussion, this suggestion was turned down on the basis that
> > recursive builds are inherently flawed and therefore, GNU make won't
> > incorporate any feature to help those who maintain such builds.
>
> I don't think that was the conclusion. I know Paul is thinking about
> implementing .WAIT or something similar. I, however, think that .WAIT
> approach is a bad way to do this. I think instead we should add another
> type of prerequisite model which further relaxes the order-only one.

Hello,

I have read again the suggested Savannah bugs, and I agree with Boris: I 
don't see any negative conclusion about this .WAIT.

As a side note, I am participating in the distmake project 
(http://distmake.sourceforge.net -- I'm not the main maintainer though),
which is basically a patch over GNU make 3.80, in order to distribute 
jobs on several machines.

I am also aware of the "Recurvise Make Considered Harmful" article 
(http://www.pcug.org.au/~millerp/rmch/recu-make-cons-harm.html), but 
it's a bit old now, and it's not always practicable to write Makefiles 
in a non recursive manner.

Anyway, when spreading make over several machines, you can easily use 
rather large values for -j, and it often exhibits errors.

A possible fix in some cases includes adding a recursive call to make 
(even with the same directory), so as to force sequentiality, but it's 
not elegant.

Alexey's proposal seemed nice to me.

Regards,

Christophe.