Carsten Neumann wrote:
> Allen Bierbaum wrote:
>
>> Carsten Neumann wrote:
>>
>>> Hi Allen
>>>
>>> Carsten Neumann wrote:
>>>
>>>
>>>> Hi Allen,
>>>>
>>>> Allen Bierbaum wrote:
>>>>
>>>>
>>>>> I just updated OpenSG2 after about a month and I ran into a strange
>>>>> issue.
>>>>>
>>>>>
>>> Do you know the revision of your previous tree ? I think the change
>>> comes from setting
>>>
>>> stop_traversal = True
>>>
>>> in the Source/System/Depreciated/build.info file in r379. Looking at the
>>> code in the current and old SConstruct files, I believe they would both
>>> show the behavior you observe.
>>>
>>> Maybe the meaning of stop_traversal is not clearly defined, should it mean:
>>>
>>> 1) ignore subdirectories of this directory, but pick up files in current dir
>>>
>>> 2) ignore this directory and everything below (the current)
>>>
>>>
>> 2 is what is it supposed to mean. I am updating the documentation in
>> that file to note the change.
>>
>
> yeah, it also makes the most sense IMHO.
> So, the System/Depreciated/build.info file should probably look like this:
>
> enable_osg1_compat = opts.GetOption("enable_osg1_compat");
>
> if enable_osg1_compat.getSettings()[0][1] == False:
> stop_traversal = True
> else:
> stop_traversal = False
>
I don't think this will work right now. You may need to update the
traverser to work in the case where no library is specified. Give it a
go and let me know what you find though.
>
>> On the topic of the build system I finally had time to review some of
>> the changes you made and I have a few comments:
>>
>> - I think we should remove the ";"'s at the end of some of the lines.
>> It is not needed for python and it is probably better to have the entire
>> file be consistent.
>>
>
> ok, will remove them.
>
>
>> - Some of the helper methods could be moved to a separate file like
>> "tools/scons-build/helpers.py" or something similar. This could make
>> the main SConstruct file a bit less intimidating.
>>
>
> agreed, will do so.
>
>
>> - There is still a lot of things that could be improved but I think most
>> of these owe to the fact that the variant and environment builder code I
>> wrote is still very rough and undocumented. I will see what I can do on
>> this with the limited time I have right now.
>>
>
> Hm, I'm not sure if it makes sense to push the env. builder to the
> limit, at some point platforms/compilers are just too different, so I
> think it would not be bad to have files that do platform specific setup.
> Though, having the windows build work would make some people quite happy
> I think ;)
>
>
I think the variants is the killer right now. I was trying to be smart
and make it possible to build 32bit and 64bit code at the same time, but
this just does not work well with the options code. I think I will
strip that part out of variants. And if that doesn't make variants work
better then I think we should just remove it entirely. It is a good
idea by the implementation ended up very nasty.
The env builder seems to work very well for me. I have used it on
several projects now and I am very happy with that part of the code.
As far as the windows build goes, I had a report this morning that it is
working on windows for atleast one person. So I think the problems on
windows now come down to differences between user environments. It
would be very helpful to have people show what their current problems
are. It would be even more helpful to get a lot more machines running
in the buildbot setup. :)
-Allen
-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
_______________________________________________
Opensg-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/opensg-users
