Re: Slow clsql compilation

Nikodemus Siivola wrote, On 23/8/06 2:44 AM:
> Probably. For stepping it depend on the policy the original function
> was compiled with -- if it had a low debug policy, then inlining it
> would actually improve the STEP-experience.
> 
> ...but for backtraces and tracing, definitely.

I noticed this in the policy section of the manual:

"(max 1 speed space compilation-speed)
    If debug is also at least 2, then the code is partially steppable. If
debug is 3, the code is fully steppable. See Single Stepping, for details.
Fully steppable code take exponentially longer to compile in some cases, and
is significantly larger and slower; for partially steppable code the speed and
space penalties are significantly smaller."

Notice what it says about exponentially longer compilation for steppable code.
It didn't say that for inline code. I tried compiling with (debug 3) (speed 1)
(space 2) to see if that would inhibit inlining and speed things up as you
said in your previous message, and the compile took almost as long. (debug 3)
(speed 1) took 193 seconds on my machine, while adding (space 2) resulted in
189 seconds.

Then I tried (debug 2) (speed 1) (space 2) and it took 2.7 seconds. I removed
the (space 2) from that and it took 0.97 seconds.

 Sidney Markowitz
 http://www.sidney.com


-------------------------------------------------------------------------
Using Tomcat but need to do more? Need to support web services, security?
Get stuff done quickly with pre-integrated technology to make your job easier
Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642