[Python-Dev] A word from the author (was "pymalloc", was "fun", was "2.1 slowe r than 2.0")

[Tim Peters]

[Vladimir Marangozov]

Hi Vladimir!  It's wonderful to see you here again.  We had baked a cake for
your return, but it's been so long I'm afraid I ate it <wink>.

Help us out a little more, briefly.  The last time you mentioned obmalloc on
Python-Dev was:

    Date: Fri, 8 Sep 2000 18:23:13 +0200 (CEST)
    Subject: [Python-Dev] 2.0 Optimization & speed
    > ...
    > The only reason I've postponed my obmalloc patch is that I
    > still haven't provided an interface which allows evaluating
    > it's impact on the mem size consumption.

Still a problem in your eyes?  In my eyes mem size was something most people
would evaluate via their system-specific process monitoring tools, and they
wouldn't believe what we said about it anyway <0.9 wink>.

Then the last thing mentioned in the patch

http://sourceforge.net/patch/?func=detailpatch&patch_id=101104&
    group_id=5470

was 2000-Aug-12 13:31:

> Status set to Postponed.
>
> Although promising, this hasn't enjoyed much user testing for the
> 2.0 time frame (partly because of the lack of an introspective
> Python interface which can't be completed in time according to
> the release schedule).

But at that time it had been tested by more Python-Dev'ers than virtually
any other patch in history (yes, I think two may still be the record <0.7
wink>), and nobody else was *asking* for an introspective interface -- they
were just timing stuff, and looking at top/wintop/whatever.

Now you seem much less hesitant, but still holding back:

> Because the risk (long-term) is kind of unknown.

I'll testify that the long-term risk of *any* patch is kind of unknown, if
that will help.

> ...
> I'd say, opt-in for 2.1. No risk, enables profiling.

Good.

> My main reservation is about thread safety from extensions (but
> this could be dealt with at a later stage)

I expect we'll have to do the dance of evaluating it with and without locks
regardless -- we keep pretending that GregS will work on free-threading
sometime *this* millennium now <wink>.

BTW, obmalloc has some competition.  Hans Boehm popped up on c.l.py last
week, transparently trying to seduce Neil Schemenauer into devoting his life
to making the BDW collector make Python's refcounting look like a cheap
Dutch trick <wink>:

    http://www.deja.com/getdoc.xp?AN=722453837&fmt=text

you-miss-so-much-when-you're-away-ly y'rs  - tim