Author vstinner
Recipients neologix, vstinner
Date 2013-11-27.18:03:10
SpamBayes Score -1.0
Marked as misclassified Yes
Message-id <1385575392.84.0.858805099832.issue19817@psf.upfronthosting.co.za>
In-reply-to
Content
It would be nice to add a memory_limit feature to the tracemalloc to make memory allocation fails if it would make the traced memory greater than the limit.

See also the pyfailmalloc project which is similar but different:
https://bitbucket.org/haypo/pyfailmalloc

pyfailmalloc doesn't count allocated bytes, it schedules an error in N allocations where N is a random number. And pyfailmalloc only schedules one error, whereas I propose to make all memory allocations fails until the limit is respected.

So if you only 0 bytes free (according to the limit), all memory allocations fail until some bytes are freed.

Python currently behaves badly under very low memory condition.

See also the issue #19437.
History
Date User Action Args
2013-11-27 18:03:12vstinnersetrecipients: + vstinner, neologix
2013-11-27 18:03:12vstinnersetmessageid: <1385575392.84.0.858805099832.issue19817@psf.upfronthosting.co.za>
2013-11-27 18:03:12vstinnerlinkissue19817 messages
2013-11-27 18:03:12vstinnercreate