Hi,
If you forward your analysis to us, we can add it to a bug for you. Surely
we can come up with a way to optimize environment handling (perhaps with a new
interface that future executables will use, while deprecating the existing
mechanism). Such pathological performance is certainly a bug and, if its
severity turns out to be as dire as you say, it needs to be fixed.
--S
Quoting ольга крыжановская, who wrote the following on Wed, 16 Jun 2010:
I already looked at the problem. This can not be fixed without
breaking binary compatibility.
Olga
2010/6/16 Darren Reed <[email protected]>:
On 15/06/10 03:31 PM, ольга крыжановская wrote:
Solaris getenv() and putenv() do not scale linear and it gets a lot
*worse* above 100 or more variables. Adding 1 environment variable per
libmtmalloc tunable makes IMO no sense, it is not economic nor is it
fast to have that many variables for similar purposes.
I don't buy the comment about server applications, this is about
economic engineering, this should apply to all users of libmtmalloc,
the small ones and the larger server applications.
Then file a bug for getenv() and putenv() (maybe even 1 each?),
stating that their performance does not scale well.
Darren
2010/6/15 Bart Smaalders<[email protected]>:
On 06/15/10 10:58, ольга крыжановская wrote:
I have a comment: Why do you use 3 environment variables? Environment
variables are very expensive to use, they significantly increase start
up time of an application and slow down all applications which use
getenv() or putenv() if the size of the environment is large. libast
had the same problem, with bitter and measurable impact on start up
time, and replaced all the allocator tunable variables into 1 variable
to improve start up time.
Sorry... this makes little sense. Yes, if you have ten thousand
environment variables, getenv can take significant time.... but this
isn't the common case, and any program that needs mt malloc isn't
worried about optimizing startup time.
Typical use case is server applications.
- Bart
--
Bart Smaalders Solaris Kernel Performance
[email protected] http://blogs.sun.com/barts
"You will contribute more with mercurial than with thunderbird."
--
, _ _ ,
{ \/`o;====- Olga Kryzhanovska -====;o`\/ }
.----'-/`-/ [email protected] \-`\-'----.
`'-..-| / http://twitter.com/fleyta \ |-..-'`
/\/\ Solaris/BSD//C/C++ programmer /\/\
`--` `--`
_______________________________________________
opensolaris-arc mailing list
[email protected]
