Microsoft compares Win2k to Linux

Björn Lindberg at
Fri Jul 19 17:13:44 PDT 2002

Tony Karakashian wrote:
> >> Really? Is this for all Windows versions? That is very broken behaviour
> >> in any case.
> You can't just arbitrarily call something "broken".  If it was a severe

Yes I can. If the behaviour is like it was described, it is broken. No
OS should fail to release all resources after a terminated process.

> issue that could simply be fixed by changing the way a call makes, I'm
> sure it would have been fixed by now.  Of course, their "fix" may be
> they assume a programmer will make the right calls.  Too often I hear
> complaints from the Linux community "MS just assumes the way they do
> things is the right way.  I should have more flexibility!"  The fix,
> in this case, is to just use the calls properly.

No. These calls apparently belongs to a higher level library. Memory
managment is a low level fundamental issue, that the kernel has to

> Let's assume that it was that simple to fix.  We'll just say that
> we modify the SysAllocString function so that when the app closes, Windows
> calls the SysFreeString function for it.  Well, then, any app that

The kernel doesn't have to call anything, it should simply free all
memory allocated by the process when it is terminated.

> specifically didn't call it for some intentioned reason would be broken and
> might crash or have some other problems.  MS then gets a lot of crap about
> changing APIs just to lock out competition....There may be reasons things
> aren't "fixed".  Such as, they may not be broken.

As I understood it, this was a specific MS API, so it is not a matter of
not locking anybody out.

A large program can have memory leaks, which will cause it to increase
in size as it is run for a longer period of time, but the OS should
always reclaim all resources when the process is terminated. If the OS
fails in it's memory management such that badly written programs can
cause memory to be allocated without ever being freed again, the only
way to free that memory is to restart the OS. That is certainly broken
behaviour, and the sign of an inherently unstable OS.

Unsubscribe: send email to listar at
and put 'unsubscribe lfs-chat' in the subject header of the message

More information about the lfs-chat mailing list