Werner LEMBERG <[email protected]> writes:
>>> In file included from parse-scm.cc:26:0:
>>> ./include/lily-lexer.hh:28:21: fatal error: parser.hh: No such
>>> file or directory
>>
>> More likely `make -j3 all'...
>
> No, it was a single threaded make call -- I always divert both stdin
> and stdout of `make all' to a file, and with -j3 this output would be
> much harder to read.
>
> BTW, in older GNU make versions, files were processed alphabetically,
> but this is no longer the case. It seems that my GNU make 3.82
> accidentally selected a processing order which triggered the problem.
Or mine one which didn't. I _did_ use make -j3 for the tests before
committing the fix just to make somewhat sure that a multijob make would
not accidentally suffer, and it might just be that this left the
single-job variant untested.
Ugh. Perhaps I should just cater for the single-job variants and let
others sort out the rest. Though -j3 actually does seem to help a bit
here ("plenty" of memory, laptop disk).
--
David Kastrup
_______________________________________________
lilypond-devel mailing list
[email protected]
https://lists.gnu.org/mailman/listinfo/lilypond-devel
