[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: gcc --coverage and copying Bison generated files
From: |
Akim Demaille |
Subject: |
Re: gcc --coverage and copying Bison generated files |
Date: |
Thu, 15 Jan 2015 11:00:32 +0100 |
> Le 27 nov. 2014 à 05:03, Daiki Ueno <address@hidden> a écrit :
>
> Hello,
Hi Daiki!
> Sorry for that this is not really related to Bison (but GCC or LCOV),
> but perhaps a common question that anyone might know the solution.
>
> In gettext we have the following directory structure:
>
> gettext-runtime/
> intl/
> Makefile
> plural.y
> plural.c # <-- generated from plural.y
> ...
>
> gettext-tools/
> intl/
> Makefile
>
> In order for gettext-runtime and gettext-tools to be built separately,
> gettext-tools/intl/Makefile refers to the source code in
> gettext-runtime/intl/ instead of copying the object files.
>
> So gettext-tools/intl/Makefile looks like:
>
> srcdir = ../../gettext-runtime/intl
>
> plural.lo: $(srcdir)/plural.c
> $(AM_V_CC)$(LIBTOOL) $(AM_V_lt) --tag=CC --mode=compile \
> $(COMPILE) $(srcdir)/plural.c
>
> This works for normal compilation. However, when compiling with the
> --coverage option of GCC, the generated .gcno file embeds relative file
> name annotated with #line directive in plural.c. That confuses LCOV:
>
> genhtml: ERROR: cannot read /tmp/...snip.../gettext-tools/intl/plural.c
>
> http://hydra.nixos.org/build/17564911/log/tail-reload
>
> Is there any easy way to tell GCC the correct source file location? Or
> should we copy all the source files?
I don't know. I face this problem constantly myself with
GCOV, not just for Bison files, but for almost any kind of
generated file I have in my various projects.
IMHO, what really lacks is a -I like option for gcov/genhtml.
If you find something better, please tell us!
[Prev in Thread] |
Current Thread |
[Next in Thread] |
- Re: gcc --coverage and copying Bison generated files,
Akim Demaille <=