1. Rob: Another small feature request for v3.0
Hey Rob,
Do you think maybe we can have 'time profiling' for Windows, Linux,
and FreeBSD Euphoria platforms for the next major release?
Pleeeaase... 
It would be a great addition; it could work just like the DOS-32 implementation.
A "performance nut" like myself would find this cool feature very valuable, as he
begins showing more interest in cross-
platform development.
The majority of Euphoria programmers write software for Windows
and/or Linux and would really welcome this addition too?
What do you think Rob?... Cool?
Does 'execution-count profiling' work on FreeBSD Euphoria?
In the documentation it says: DOS, Windows, and Linux.
Regards,
Vincent
2. Re: Rob: Another small feature request for v3.0
Vincent wrote:
> Hey Rob,
>
> Do you think maybe we can have 'time profiling' for Windows, Linux,
> and FreeBSD Euphoria platforms for the next major release?
>
> Pleeeaase...
>
> It would be a great addition;
<snip>
I would also appreciate it.
Regards,
Juergen
3. Re: Rob: Another small feature request for v3.0
- Posted by Robert Craig <rds at RapidEuphoria.com>
Nov 19, 2005
-
Last edited Nov 20, 2005
Vincent wrote:
> Do you think maybe we can have 'time profiling' for Windows, Linux,
> and FreeBSD Euphoria platforms for the next major release?
>
> Pleeeaase...
>
> It would be a great addition; it could work just like the DOS-32
> implementation.
> A "performance nut" like myself would find this cool feature very valuable,
> as he begins showing more interest in cross-
> platform development.
>
> The majority of Euphoria programmers write software for Windows
> and/or Linux and would really welcome this addition too?
>
> What do you think Rob?... Cool?
I'd like to have that too.
There was a thread on this a year ago:
http://www.listfilter.com/cgi-bin/esearch.exu?thread=1&fromMonth=9&fromYear=9&toMonth=B&toYear=9&keywords=%22profile_time+in+win32+euphoria%22
A clever person could add a form of time profiling to the
PD source (execute.e). Just record the time() when control
enters, and the time() when it exits each routine. Then you
could say how much time was spent inside each routine.
The difference in time() values would often be zero, but
it would average out to a nonzero value if the routine was called
many times. I've done this before for C code, and it worked well.
Of course this isn't as desirable as having it at the statement level
in the official interpreter.
> Does 'execution-count profiling' work on FreeBSD Euphoria?
Yes.
> In the documentation it says: DOS, Windows, and Linux.
I'll fix that.
Regards,
Rob Craig
Rapid Deployment Software
http://www.RapidEuphoria.com
4. Re: Rob: Another small feature request for v3.0
- Posted by Vincent <darkvincentdude at yahoo.com>
Nov 19, 2005
-
Last edited Nov 20, 2005
Robert Craig wrote:
> I'd like to have that too.
> There was a thread on this a year ago:
> <a
> href="http://www.listfilter.com/cgi-bin/esearch.exu?thread=1&fromMonth=9&fromYear=9&toMonth=B&toYear=9&keywords=%22profile_time+in+win32+euphoria%22">http://www.listfilter.com/cgi-bin/esearch.exu?thread=1&fromMonth=9&fromYear=9&toMonth=B&toYear=9&keywords=%22profile_time+in+win32+euphoria%22</a>
>
> A clever person could add a form of time profiling to the
> PD source (execute.e). Just record the time() when control
> enters, and the time() when it exits each routine. Then you
> could say how much time was spent inside each routine.
> The difference in time() values would often be zero, but
> it would average out to a nonzero value if the routine was called
> many times. I've done this before for C code, and it worked well.
>
> Of course this isn't as desirable as having it at the statement level
> in the official interpreter.
>
I see, but there is no profiling features in the PD-source and only
trace type 3 is supported. I'm talking about implementing it in the official
interpreter C backend. I believe there is some Win 32 APIs
for measuring time, but I'm not sure about Linux or FreeBSD. There
are a few machine coded solutions in the archive, but they seem to
require Windows or DOS plus a Pentium class machine; that probably
will not cut it though on older 386-486 machines and Linux/FreeBSD.
If you can't figure out a solution for the other three platforms,
perhaps you can just use time() like you say. It might not be the
most accurate, but it should be good enough; you can just keep the
interrupt timer used for DOS Euphoria the way it is.
Perhaps someday you'll figure out how to improve time()'s resolution,
but if not, thats ok too.
> > Does 'execution-count profiling' work on FreeBSD Euphoria?
>
> Yes.
>
> > In the documentation it says: DOS, Windows, and Linux.
>
> I'll fix that.
>
Okay... thanks.
> Regards,
> Rob Craig
> Rapid Deployment Software
> <a href="http://www.RapidEuphoria.com">http://www.RapidEuphoria.com</a>
Regards,
Vincent
