Total Self Cost / Total Inclusive Cost show 0

[Toast]

Hello. I stumbled upon webgrind today, and I really like it's
simplicity. I was able to install xdebug and webgrind in under 30
minutes. FYI, I'm using 5.3.2 on ubuntu 10.04, xdebug 2.1.0.

Webgrind is working, as I can see a generated cachegrind file, my
custom functions, invocation count, links to files, etc. However, all
cost metrics are reading as 0.00. Also, it's returning a null value
for milliseconds, ie this is displayed towards the top right:
- 136 different functions called in milliseconds ( runs, 136 shown)


I see that development on this project has really slowed to a crawl;
is anyone still working to maintain it?

Thanks!

[Joakim Nygård]

Glad you found webgrind and like it. Indeed development has been slow
lately. We are considering moving webgrind to github to allow the
community better access and ensure future development, regardless of
our available time.

[oetting]

Regarding you problems with using webgrind.

It could sound like this issue: http://code.google.com/p/webgrind/issues/detail?id=60&can=1

Do you known if your cachegrind file is missing the summary line? (a
line starting with "summary:")

On Jan 3, 6:36 pm, Toast <noh...@gmail.com> wrote:

[Toast]

Grepping through the cachegrind file shows no summary line.  I'm assuming it's the one saved in tmp.

The issue sounds exactly the same.  We do use exit within the code, but it should only be triggered when an error occurs.  The page is rendering correctly, so I do not believe exit is being called.

It looks like the second xdebug bug was marked resolved in 2007 (http://bugs.xdebug.org/view.php?id=631).  That'd be crazy if it popped up a 3rd time. =D

[oetting]

On Jan 3, 9:25 pm, Toast <noh...@gmail.com> wrote:
> Grepping through the cachegrind file shows no summary line.  I'm assuming
> it's the one saved in tmp.
>

The file created by xdebug. Typically called something like
cachegrindXXX.out. That file should contain a "summary: X" line for
webgrind to work.

> The issue sounds exactly the same.  We do use exit within the code, but it
> should only be triggered when an error occurs.  The page is rendering
> correctly, so I do not believe exit is being called.

Ok. Strange.

>
> It looks like the second xdebug bug was marked resolved in 2007 (http://bugs.xdebug.org/view.php?id=631).  That'd be crazy if it popped up a
> 3rd time. =D

That was the issue we reported. Please look at the date again, it is
from 2010 :)

[Toast]

The file is cachegrind.out.17292, which is the same one webgrind refers to in it's drop-down.  It does not contain a summary line.

My bad, definitely read the date backwards. *facepalm*

There's no rush on this.  I'm (slowly) trying to automate some of my more routine tasks.  Being able to run an automated performance check of our site before deploy would be nice.

[Olle Jonsson]

Dear Toast and webgrinders,

The Automated Performance Check concept sounds like WIN.

Could you lay out your assumptions, and steps to get there? Perhaps
we, the larger group, could assist in some way.

[Toast]

I have a realistic plan and an ideal plan:

Realistic:  Turn webgrind on before one of our testers does a full site sweep, have him note any pages that load exceptionally slow, and analyze the performance afterwards.  This is automated in that I don't have to do it. =D

Ideal:  Set up selenium (or something else from http://openqa.org/) in conjunction with webgrind.  Schedule a full site sweep to run every night, and review the results in the morning.  Possibly pull/graph these results on an ongoing basis.  Since I haven't actually gotten around to setting up selenium yet, I'm not sure how viable this plan is.

Of course, step one is getting this bug figured out. =D

[Nicolas Mendoza]

This happens also when a script ends because it runs too long. How can I fake a summary line for now?