On Fri, 21 Mar 2003, Stas Bekman wrote: > Date: Fri, 21 Mar 2003 14:09:11 +1100 > From: Stas Bekman <[EMAIL PROTECTED]> > To: Jie Gao <[EMAIL PROTECTED]> > Cc: [EMAIL PROTECTED] > Subject: Re: cvs version make test errors > > Jie Gao wrote: > > Hi All, > > > > I got a version from cvs today and make test fails with: > > > special_blocks.....ok 5/12unable to find interp > > de76d477-58b9-0310-b94d-dd991812e62e > > Thanks Jie. > > Your bug report missing the mod_perl and apache information. You should have > run t/REPORT to get this as explained here: > http://perl.apache.org/docs/2.0/user/help/help.html#Reporting_Problems > I can't tell whether you are running a threaded-mpm or a prefork. > > However, this is not a bug in mod_perl, but a current limitation in one of the > testing framework functionalities. Certain tests need to make sure that they > hit the same interpreter twice in a row (e.g. to test the closure effect), > apparently the current implementation doesn't work well on several platforms, > win32 included. > > If somebody (preferrably on a platform that has this problem) can look at this > issue and resolve it, that would be a great help. > > You can find the code in Apache-Test/lib/Apache/TestRequest.pm and you have > several tests that use it (grep for 'same_interp_tie'). The currently used > algorithm is very simple: the first request stores the unique id of the perl > interpreter in the response headers and on the following requests the client > supplies this header. Apache::TestRequest resubmits the request again and > again till it hits the same interpreter and only then returns the response to > the client. After TRY_TIMES (50) it'll give up and return an error. I'm not > sure whether raising of this number to let's say 500 will help to solve the > problem. > > Your help is appreciated.
Guess what, I ran the test again, it passed. :-) I figure this had to do with the load of the machine. Anyway, I was using threaded-mpm with apache 2.0.44. Many thanks, Jie
