Joe Neeman wrote:
> It seems that some changes in the way lilypond-books output is stored
> have broken certain build rules, particularly ones that clean stuff.
> For example,
>
> $ make check
> $ make test-clean
> $ time make check
> real 0m31.410s
> user 0m25.598s
> sys 0m4.084s
> $ make test-clean && rm -fr out/lybook-db
> $ time make check
> real 6m42.076s
> user 6m18.948s
> sys 0m14.981s
>
> Since the first build time is so short, it seems that 'make
> test-clean' is not correctly cleaning everything. I think there is a
> similar problem with make web-clean, but I haven't investigated
> thoroughly because 'make web' takes so long anyway.
Incomplete test-clean and web-clean can be cumbersome when you're
working on C++ code, but it's very useful when you don't want to rebuild
all lp-book snippets, e.g. when you edit the documentation or hack
documentation building.
If you don't want to rebuild all of the snippets, why do 'make web-clean' in the first place?
make clean also cleans lp-book snippets, but maybe it's worth to make a
new top-level target snippets-clean that would do 'rm -rf out/lybook-db'
and would be called by test-clean?
This would be sufficient for me, since I use test-clean much more often than web-clean. I'd rather not have to make clean every time I need to see if I've caused a regression.
Cheers,
Joe