[xdebug-general] Re: feature request - dump traces file to disk only if total script execution time is greater or equals a threshold in seconds

Derick, thanks for the explanation.

Deleting files is ugly indeed from the architectural point of view, but on
the other hand the result would greatly speed up investigation process.

Probably you can suggest a simpler way to filter out traces files that were
processed longer than X seconds? Maybe using grep/awk from shell?

---
WBR, Valeriy
On Tue, Nov 18, 2014 at 2:43 PM, Derick Rethans <derick[@]xdebug.org> wrote:
> Hi!
>
> On Mon, 3 Nov 2014, Valeriy wrote:
>
> > Normally, if one is not satisfied with load/stress testing results and
> > wants to investigate root causes, only requests that took longer than
> > X seconds to execute matter. However, right now such testing will
> > generate tons of profiler traces files, so additional manual files
> > cleaning/filtering has to be done.
> >
> > It would be great to have a configuration parameter like the
> > following:
> >
> > xdebug.profiler_request_time_threshold
> > Type: integer, Default value: 0
> >
> > When this setting is set to a value greater than 0, trace files will
> > be dumped to disk only if total request processing time is greater or
> > equals the threshold (in seconds).
>
> Right now, Xdebug writes trace files/profile files on the fly as to
> minimize the amount of memory being used. As trace files and profile
> files can be really large (100s of MBs), you'll see that I will have to
> do that.
>
> Of course, I could delete these files when at the end of a request the
> threshold time has not been reached, but that's a bit ugly.
>
> cheers,
> Derick
>
> --
> Like Xdebug? Consider a donation: http://xdebug.org/donate.php
> Xdebug | http://xdebug.org | xdebug-general@lists.xdebug.org
> twitter: @derickr and @xdebug
>