On Tue, 2008-04-15 at 11:14 -0700, Bernard Li wrote: > Hi Carlo: > > On Tue, Apr 15, 2008 at 1:11 AM, Carlo Marcelo Arenas Belon > <[EMAIL PROTECTED]> wrote: > > > since that code has been overridden by the one in r1230, is this still > > relevant? > > The original patch which Matt submitted (and which Brad checked in) > had a bug which simply does not set the cookie if it exists. This is > simply incorrect because the content of the cookie might have changed > and this was exactly why gridstack wasn't working. This was fixed in > the backport that is r1230. > > However, the original idea from Matt which is to conditionally set the > cookie is still there, it's just more strict. And this is the issue > we are discussing here -- how to determine the situation when the > cookie should be set, or should we just let the cookie be set all the > time. > > > isn't setting the cookie all the time (as you proposed), going to trigger > > the > > bug why the fix in r790 was introduced to begin with? > > Yes, it would. But I don't see any way to solve both issues. I > simply don't see this exponential growth of cookie using a typical > browser like Firefox, so I am trying to get a better understanding of > what this library Matt is working with is doing. Otherwise, reverting > the original patch and simply set cookie all the time seems like the > quickest fix. > > Brad, can you please get a hold of Matt to see if he has any input?
At the time we were embedding the Mozilla libraries in an Eclipse RCP application - not Firefox; it was Seamonkey or some other dynamically-linkable Mozilla library (Firefox libs are statically linked on SUSE at least, at least they were back then). IIRC this is where I was seeing the behavior in question. I don't know if it can be reproduced, if more current versions of these libraries still have this problem, etc. I don't exactly remember the details, sorry (I should have filed a bug, I apologize). I'm not on that project (the Eclipse RCP app) anymore, and I don't think there are plans to continue to access the monitoring data via a web browser - I think they will get it via different means moving forward. It *may* be that we were the only ones seeing the problem. My $0.02 - revert back to what was working before. If we see a problem, perhaps we can do a user-agent check at that level or something else to work around it, even if we (Novell) just do it in a patch in our version of the RPM, if it turns out that we are the only ones seeing the problem. BTW, sorry for not replying earlier - I can never tell if you are referring to me or Matt Massie. :) ------------------------------------------------------------------------- This SF.net email is sponsored by the 2008 JavaOne(SM) Conference Don't miss this year's exciting event. There's still time to save $100. Use priority code J8TL2D2. http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone _______________________________________________ Ganglia-developers mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/ganglia-developers
