On 03/12/14 16:14, Samuel Iglesias Gonsálvez wrote:
> On Wednesday, December 03, 2014 03:44:19 PM Samuel Iglesias Gonsálvez wrote:
>> On Wednesday, December 03, 2014 03:29:46 PM Samuel Iglesias Gonsálvez wrote:
>>> On Wednesday, December 03, 2014 11:34:10 AM Emil Velikov wrote:
>>>> On 03/12/14 11:30, Emil Velikov wrote:
>>>>> Hi Iago,
>>>>>
>>>>> On 01/12/14 11:04, Iago Toral Quiroga wrote:
>>>>>> From: Jason Ekstrand <jason.ekstr...@intel.com>
>>>>>>
>>>>>> We were auto-generating it before. The problem was that the
>>>>>> autogeneration
>>>>>> tool we were using was called "copy, paste, and edit". Let's use a
>>>>>> more
>>>>>> sensible solution.
>>>>>>
>>>>>> Signed-off-by: Jason Ekstrand <jason.ekstr...@intel.com>
>>>>>>
>>>>>> v2 by Samuel Iglesias <sigles...@igalia.com>
>>>>>> - Remove format_pack.c as it is now autogenerated
>>>>>> - Add usage of INDENT_FLAGS in Makefile.am
>>>>>> - Remove trailing blank line
>>>>>>
>>>>>> v3 by Samuel Iglesias <sigles...@igalia.com>
>>>>>> - Merge format_convert.py into format_parser.py
>>>>>>
>>>>>> - Adapt pack_*_* function generations
>>>>>>
>>>>>> - Fix out-of-tree build
>>>>>>
>>>>>> Signed-off-by: Samuel Iglesias Gonsalvez <sigles...@igalia.com>
>>>>>> ---
>>>>>>
>>>>>> src/mesa/Makefile.am | 14 +
>>>>>> src/mesa/Makefile.sources | 2 +-
>>>>>> src/mesa/main/format_pack.c | 2982
>>>>>> --------------------------------------
>>>>>> src/mesa/main/format_pack.c.mako
>>>>>>
>>>>>> | 897 ++++++++++++
>>>>>>
>>>>>> src/mesa/main/format_parser.py | 71 +
>>>>>> src/mesa/main/run_mako.py | 7 +
>>>>>> 6 files changed, 990 insertions(+), 2983 deletions(-)
>>>>>> delete mode 100644 src/mesa/main/format_pack.c
>>>>>> create mode 100644 src/mesa/main/format_pack.c.mako
>>>>>> create mode 100644 src/mesa/main/run_mako.py
>>>>>>
>>>>>> diff --git a/src/mesa/Makefile.am b/src/mesa/Makefile.am
>>>>>> index 932db4f..849c9da 100644
>>>>>> --- a/src/mesa/Makefile.am
>>>>>> +++ b/src/mesa/Makefile.am
>>>>>> @@ -19,6 +19,8 @@
>>>>>>
>>>>>> # FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
>>>>>> OTHER
>>>>>> DEALINGS # IN THE SOFTWARE.
>>>>>>
>>>>>> +BUILDDIR_ABSOLUTE_PATH = $(PWD)
>>>>>> +
>>>>>
>>>>> You should not need this new variable. Additionally I'm suspecting
>>>>> this
>>>>> is the reason between the build hickups you/Jason are seeing.
>>>>>
>>>>>> SUBDIRS = . main/tests
>>>>>>
>>>>>> if HAVE_X11_DRIVER
>>>>>>
>>>>>> @@ -66,6 +68,7 @@ BUILT_SOURCES = \
>>>>>>
>>>>>> main/get_hash.h \
>>>>>>
>>>>>> main/format_info.c \
>>>>>>
>>>>>> $(BUILDDIR)main/git_sha1.h \
>>>>>>
>>>>>> + $(BUILDDIR)main/format_pack.c \
>>>>>>
>>>>>> $(BUILDDIR)program/program_parse.tab.c \
>>>>>> $(BUILDDIR)program/lex.yy.c
>>>>>>
>>>>>> CLEANFILES = \
>>>>>>
>>>>>> @@ -89,6 +92,17 @@ main/format_info.c: main/formats.csv
>>>>>>
>>>>>> \>>
>>>>>>
>>>>>> $< > $@.tmp;
>>>>>> \
>>>>>>
>>>>>> mv $@.tmp $@;
>>>>>>
>>>>>> +$(BUILDDIR)main/format_pack.c: main/format_pack.c.mako
>>>>>> main/formats.csv
>>>>>> \
>>>>>> + main/run_mako.py
>>>>>> main/format_parser.py
>>>>>> + $(AM_V_GEN)set -e;
>>>>>> \
>>>>>> + cd $(srcdir);
>>>>>> \
>>>>>> + $(PYTHON2) $(PYTHON_FLAGS) main/run_mako.py
>>>>>> \
>>>>>> + main/format_pack.c.mako main/formats.csv >
>>>>>> \
>>>>>> + $(BUILDDIR_ABSOLUTE_PATH)/$@.tmp;
>>>>>> \
>>>>>> + cd $(BUILDDIR_ABSOLUTE_PATH);
>>>>>> \
>>>>>> + cat $@.tmp | $(INDENT) $(INDENT_FLAGS) > $@;
>>>>>> \
>>>>>> + rm $@.tmp;
>>>>>> +
>>>>>
>>>>> Can you prefix the files in the srcdir, and drop the tmp file (pipe
>>>>> directly into indent). It will make the above ~1/2 the size and a bit
>>>>> easier to follow :)
>>>>
>>>> Completely forgot, please add a couple of mines into the scons build
>>>> while you're here. src/mapi/shared-glapi/SConscript could serve as an
>>>> example.
>>>
>>> This is my first time modifying a SConscript file. Looking at what was
>>> done
>>> for format_info.c, I think that the following modification is enough to
>>> have format_pack.c autogenerated:
>>>
>>> format_pack = env.CodeGenerate(
>>>
>>> target = 'main/format_pack.c',
>>> script = 'main/run_mako.py',
>>> template = 'main/format_pack.c.mako',
>>> source = 'main/formats.csv',
>>> command = python_cmd + ' $SCRIPT ' + ' $TEMPLATE $SOURCE | ' +
>>>
>>> '$INDENT $INDENT_FLAGS > $TARGET'
>>>
>>> )
>>>
>>> I did a 80-char line wrap (it was not my email client), I suppose this is
>>> valid in SConscript, right? scons -f SConscript did not complain about it.
>>
>> Comparing with the SConscript you told me, I did a couple of small
>> modifications to my previous code:
>>
>> format_pack = env.CodeGenerate(
>> target = 'main/format_pack.c',
>> script = 'main/run_mako.py',
>> source = 'main/format_pack.c.mako main/formats.csv',
>> command = python_cmd + ' $SCRIPT ' + ' $SOURCE | ' + \
>> '$INDENT $INDENT_FLAGS > $TARGET'
>> )
>>
>> I think this is more or less what we need but, if someone with more
>> experience on SConscript files sees a better way, it would be great to
>> learn how to do it.
>>
>> I'm going to do some tests to check that this is building properly.
>>
>
> I finally found a way of generating format_pack.c and format_unpack.c with
> scons using env.CodeGenerate().
>
> As env.CodeGenerate() only uses one 'source' argument (it fails if you run
> what I wrote above), I ended up renaming format_pack.c.mako and
> format_unpack.c.mako to be python scripts with the template content inline.
>
> Now, they auto-generate format_{pack,unpack}.c files by themselves using the
> python mako module. Thanks to that, run_mako.py can be removed and both
> SConscript and Makefile.am modifications are simpler.
>
> Any other idea/suggestion would be helpful.
>
Nicely done Sam. I realise that dealing with autotools/scons can be a
pain, but it's a one-off job. This way new and old devs/users won't end
up looking for a shaman just to get it compiling :P
With the comments for 14 & 15 addressed, I think that we're good on the
build side of things :)
Thanks
Emil
P.S. "Crazy" android people can update their build as they hit it :P
_______________________________________________
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/mesa-dev
