[Scons-dev] please try latest default branch

[Kenny, Jason L]

So to answer the question Yes we do.  In fact we have two cases:

1)      Header generation ( this is done via the Parts call to PythonScript() which is a builder I added to call a python file given two functions, one for emitting what it will do and one to do the work). This is not an issue as the finial signature have the values of the header files when it is stored. As I understand it the scons sees that he header file it needs to scan is not built yet, and as such it builds it so everyone is happy.

2)      We build python and generate pyc file from that version of python we build ( yes python is part of Vtune). This is one area in which hiccups happen, I believe this is because I have not made a better pyc builder vs the hack in our build files, and I have not handle the notion passing builder from different component yet. However this was mostly corrected in our system.. so this is not what is rebuilding for me at least.

What I am seeing is that Scons reports the build sig is different, however we seem to have the same data in the same order in one case, in other a value is swapping around. The latter is a bug in Parts mapping code ( caused by issue in the subst engine in Scons I have not worked around). The first case looks like SCons, but I am not sure of what it is yet. I have generally ignored it as it is really hard to debug, and it impossible to reproduce so far on a smaller sample, and at the moment Intel is not paying me to work on Parts, so I have not had the time to figure out what is up. I was more of a view that refactoring some code in Parts and in Scons might make it go away.

I have not tested Parts with the latest 2.3.x as of yet. My main worry is that I have hack the node object in scons via monkey patching to correct issues or add functionality ( such as symlinks), the moving to slots could break and fixes we have made in Parts ( a risk I knew might happen, as is mine to deal with)

Jason
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://two.pairlist.net/pipermail/scons-dev/attachments/20140110/5a74b98d/attachment-0001.htm>