[L2Ork-dev] purr data gitlab hard reset

[Jonathan Wilkes]

Hi List,I went ahead and did a hard reset to 2.2.1 in the gitlab repository. I had 
initially thought that pd-cyclone was ready to merge, but it looks like there 
are still outstanding problems with namespace prefixes and some of the 
help patches bugs to be fixed (the ones for the cyclops classes). Both 
have been reported to the pd-cyclone issue tracker:
https://github.com/porres/pd-cyclone/issues

A hard reset should not be necessary in the normal course of development. 
Probably we should have tested out the CI binaries from the "update-cyclone" 
feature branch, rather than merging them first and then testing. I'll try to keep 
that in mind for the future.
I've got some soundfiler fixes ported from Pd Vanilla, so I'll try to merge those 
for a bugfix release in the next few days.

There's also a big memory inefficiency in Purr Data/Pd-l2ork's due to changes 
to Pd's memory allocator in m_memory.c. Essentially, "getbytes" and friends 
are multiplying the request bytes by 4 or 8, depending on what the sizeof(char*) 
is on that particular system.
The commit message for the getbytes change is this:
    *cleaned up the way rtext memory management is handled to silence valgrind errors

It appears Ico made the change.
That seems very fishy to me, and it's been noticed by a user whose tables are 
ballooning by 8x when compared to Pd Vanilla.
Anyone have a clue why the memory allocator would be changed to accommodate 
rtext memory problems? Or, what to do now that the bugfix is causing unnecessary 
memory overhead for a user?

Thanks,Jonathan
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://disis.music.vt.edu/pipermail/l2ork-dev/attachments/20170618/cba3cf27/attachment.html>