Re: [Evolution-hackers] Memory testing > 15,000 headers

[Philip Van Hoof <spam pvanhoof be>]

On Tue, 2006-08-22 at 16:54 +0200, Philip Van Hoof wrote:

> You can repeat this testing using the tests in the tinymail framework:

Also make sure you read this page:

http://tinymail.org/trac/tinymail/wiki/HelpMemoryTesting

Don't forget to perform G_SLICE=always-malloc and G_DEBUG=gc-friendly if
you want to check for leaks. I tested it quite some times and it doesn't
look like tinymail nor Camels IMAP provider are leaking memory in their
crucial memory-eating folder-summary related parts.

The NNTP provider, however, definitely *is* leaking here. I haven't yet
tested the other providers. I will of course investigate and try to spot
the leak soon.



My Post Scriptum reasoning for the mmap crap and these measurements:

I know this sounds like yet another round of tinymail "hey ho hey ho,
hey look hey ho!"-E-mailing.

Nevertheless, tinymail can be very useful for testing Camel and more
specifically Camels memory usage. I do recommended Evolution developers
to use tinymail for getting the memory usage of Camel down. By just the
numbers that I'm measuring and calculating, it's no surprise to me that
Evolution consumes as much memory as it does.

Also try this valgrind hack: http://pvanhoof.be/files/project1.tar.gz

However, I do believe lots of improvements can be made. Sadly most of
the improvements mean irreversible and hard to make design decisions and
changes to Camel that wont be backward compatible.

I'm indeed posting these results in the hopes that people someday find
the courage and reasoning for starting the Disk Summary branch. I'm
posting in the hopes that when people will be redesigning Camel (or
libspruce), that they will understand where exactly memory is being
consumed *today*.

I indeed ALSO want to get memory consumption per active/open folder
down. The less memory one such folder consumes, the larger the folder
can be. The lesser memory the mobile device vendor must install. The
cheaper the per-unit factory cost will be and more importantly from a
technical point of view: the fewer power it will consume.

The more interesting tinymail (and therefore also Camel) will be for
mobile purposes.

> svn co https://svn.tinymail.org/svn/tinymail && cd tinymail/trunk
> ./autogen.sh --prefix=/opt/tinymail --enable-tests --with-platform=gpe
> make && make install
> 
> # Getting the headers locally (this will work out of the box, the
> # account information is in the source code indeed)
> /opt/tinymail/bin/memory-test --online
> 
> valgrind --tool=massif /opt/tiynymail/bin/memory-test -c /tmp/tinymail.0
> 
> You will mostly be using this code for this specific test:
> 
> http://svn.tinymail.org/svn/tinymail/trunk/tests/memory/memory-test.c


-- 
Philip Van Hoof, software developer at x-tend 
home: me at pvanhoof dot be 
gnome: pvanhoof at gnome dot org 
work: vanhoof at x-tend dot be 
http://www.pvanhoof.be - http://www.x-tend.be