Skip to main content

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
[cdt-dev] Minor bug (?) in make output parsing

Well, those of you who watched my earlier mail go by know that I was
having problems getting a proper AST out of a build. It turns out that
it is at least partly a make scanner problem.

All of our compile lines take the form:

  /usr/bin/ccache /coyotos/host/bin/i386-unknown-coyotos-gcc ...args...

It seems that the make output scanner doesn't know about ccache, and
therefore cannot parse this. No real surprise, I suppose, since this
usage has to be pretty rare.

I tweaked our makefiles to disable use of CCACHE, and I'm now getting
the AST that I expected.

Should the failure to handle ccache be considered a bug? If so, I'll
file one, and probably supply a patch along with it.

Curiously, now that I have a good AST I still am *not* getting a call
graph that reflects post-preprocessor code, even with full indexing
turned on. A procedure call that is hidden within a macro expansion does
not appear to be visible to the call graph generator. Is this expected,
or is this also something that should be filed as a bug?


Back to the top