This is the mail archive of the
cygwin@cygwin.com
mailing list for the Cygwin project.
Re: Win2k and cygwin memory leak
Brian.Kelly@Empireblue.com wrote:
Well, I can't feel too guilty about chiming "me too" - cause it
already brought forth a VERY useful and *productive* response:
http://cygwin.com/ml/cygwin/2003-08/msg00460.html
Unlike the ... uh-hem ... *posts* of some other folks ....
This seems to me to be just a bandaid for a specific OS (not even one
that you're running).
I will say this - anything that ends in "contact Microsoft" is about
as useful as a 300 baud modem on a Pentium 4. No - the modem is
definitely more useful.
Hey it's tough, but what else can you do really if the OS hangs on to
the memory? Bandaids like tweaking ini files and RAMpage are just that -
bandaids.
All I want to do, is use Cygwin to solve some of my problems. So I'm
given a naked Compaq box with dual processors, 2 gigs of ram, and W2K
Server w SP3. ( I am not part of the "NT Team" at work - talking to
them is a lot like talking to Microsoft. ) I set up and run nothing
but cygwin and cygwin installed apps. And then I run it again, and
again and again and again - and then the box crashes.
Again, why not install Linux?
And the NT people get all upset cause someone has to find the box in
the unlabeled server farm and power it down and back up again. Five
meetings full of name calling and finger pointing follow.
Just a slice of my "clueless" life.
But I digress.
All I want is for software ( I didn't say *cygwin* - I'm being
*generic* ) to work as "expected".
It's not "expected" for an application program to cleanup after an OS.
Crashing servers somehow violate reasonable expectations.
It's not Cygwin that crashes the system - It's Windows.
There's an awful lot of other software out there that runs 24-7 on the
same windows that you wish to blame and *it* doesn't bring the box
down. Non-cygwin Apache comes to mind. I believe it's possible to
write code that doesn't as cfg put it "triggers a windows problem"
Cause it certainly appears to me that others must have encountered the
same problem, and didn't say "well it's Microsoft's problem".
You admittedly run processes that take up huge amounts of memory and
then exit. It is the responsibility of the OS to free those resources
when the process exits. Windows still has many acknowledged memory leaks
in such situations.
Apache, OTOH, is a SERVER, designed and implement to run continually in
the background.
Why don't you look at your own "cygwin" code and implement it as a deamon?
Now I'm not telling anyone what to do, or not to do. All I know is the
Microsoft installed base probably numbers in the *billions* out there.
And even if MS *were* to fix the problem, what are the odds that this
fix would find it's way onto even a sizable fraction of that base??
Depends on how many people insist and running 5 year old OSes (like Win 98)!
In spite of the bloated bombastic verbage often spewing forth from
this forum, a *cygwin* fix is
definitely the path of least resistance.
And totally undoable. How many times must people tell you, when the
application exits there is no possible way that Cygwin can do anything
about it. Cygwin is not in the "business" like RAMPage.
Believe it or not. Someone will fix this someday. I have "faith". It
won't be me, and it won't be someone in Redmond Washington. I have a
work-around so I'm currently satisfied. I'll be patient.
Cluelessly yours,
BK
( PS - I HATE Windows - put I get paid 'quite well' to work on it, and
given the current state of the economy - I think I'll keep working on
it ..... )
I don't HATE Windows - I just understand it's limintation (and
strengths). Cygwin makes Windows a lot more bearable and usable for me.
--
Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple
Problem reports: http://cygwin.com/problems.html
Documentation: http://cygwin.com/docs.html
FAQ: http://cygwin.com/faq/