[CM] Freeing up s7 instances & memory

[Iain Duncan]

>
>
> I do hope Bil gets to merge this (or some other version of it) cause
> maintaining changes & merging the upstream.. ughh..
>

Presumably this wouldn't be too bad if you used Woody's git mirror and
rebased it over your own git branch? We used that approach at my old job
and it the nice thing is rebasing the moving master into your local branch
means the merge (and any conflicts) is always in the local branch and it's
easier to back out of. This may be old hat to you, but figured I'd mention
it case it isn't. :-)

We did
- cut feature branch, work away privately while master gets changed by
others
- when done, rebase updated master on to local branch, so your local
changes get replayed over the new master
- then merge local branch into master, which is really just cosmetic as
they are identical at that point

Git can be a pain sometimes, but man, it's good at merges compared to SVN.
Freaking magical.

iain



>
> Oh and found your original post in the mailing list:
> https://www.mail-archive.com/cmdist@ccrma.stanford.edu/msg05320.html
> (bonus: the list is searchable through mail-archive! nice thing it became
> public)
>
> On Tue, 8 Sep 2020 at 15:18, Woody Douglass <
> wdouglass at carnegierobotics.com> wrote:
>
>> Christos,
>>
>> Here's a copy of the patch i put on this list back in May. It's against
>> svn revision 804. I've been running that version since mid-May.
>>
>> Good luck!
>> Woody Douglass
>>
>>
>>
>> On 2020-09-08 09:12:27-04:00 cmdist-bounces at ccrma.Stanford.EDU wrote:
>>
>> Indeed the instances could be reused, but that would imply
>> using a common mutex amongst the running threads for any
>> s7 operation (right?).
>> What I need in my case is truly parallel instances doing their own thing.
>> I'm glad to hear that Woody has some sort of a solution.
>> I hope it gets merged at some point, but I also hope for a solution
>> in the future that frees not most but *all* of s7's memory :)
>> Any chance I could see this code? By curiosity and to see if I could
>> slowly get some understanding of the inner workings of s7.
>>
>> On Tue, 8 Sep 2020 at 01:06, <bil at ccrma.stanford.edu> wrote:
>>
>>> You're right that free(sc) does not free much.  Woody Douglass
>>> sent me code to free most of s7's memory -- the main thing
>>> is the heap -- but I haven't tackled merging it into my
>>> version of s7 yet.  I think s7 instances can be reused --
>>> just keep all your stuff in a sublet of rootlet, then
>>> set it to #f, but it would be better to actually clean up.
>>>
>>
>> _______________________________________________
> Cmdist mailing list
> Cmdist at ccrma.stanford.edu
> https://cm-mail.stanford.edu/mailman/listinfo/cmdist
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://cm-mail.stanford.edu/pipermail/cmdist/attachments/20200909/16407539/attachment.html>