Re: convert-ly question

[David Kastrup]

Martin Tarenskeen <address@hidden> writes:

> On Sun, 18 May 2014, Graham King wrote:
>
>>  On Sun, 18 May 2014 08:38:09 +0200 (CEST) Martin Tarenskeen wrote:
>> >  Luckily I don't need this kind of commandline virtuosity. I
>>        think I
>> >  can do what I need with one of the first and easiest
>>        suggestions
>> > >  convert-ly -e **/*.ly
>>
>>  One more tidbit of painfully-gained experience in this area:  If using a
>>  solution that walks the directory tree, starting convert-ly processes, it's
>>  important to do it in a way that limits the number of concurrent invocations
>>  of convert-ly.  I've managed to wedge OSX by using a tool that failed to do
>>  that :(
>
> Would this be safer?
>
> ls **/*.ly | while read f; do convert.ly -e "${f}"; done

No.  convert-ly -e **/*.ly starts just one process anyway.

> Which leads me to another Linux commandline topic:
>
> Which would be faster, on a machine with a multicore cpu and enough RAM?
>
> lilypond a.ly b.ly c.ly d.ly
> lilypond a.ly & lilypond b.ly & lilypond c.ly & lilypond d.ly
> lilypond a.ly && lilypond b.ly && lilypond.c && lilypond d.ly
>
> Thinking of it, I can test this myself.

lilypond -djob-count=4 a.ly b.ly c.ly d.ly

A good rule of thumb is one more job than CPU cores.  That should be
enough to saturate.

-- 
David Kastrup