apr-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "William A. Rowe, Jr." <wr...@rowe-clan.net>
Subject Re: Test case errors detected in APR
Date Mon, 21 Nov 2005 05:11:14 GMT
Raythe Blooandsekzy wrote:
> After downloading and compiling APR-1.2.2 on a Win2000Pro (5.00.2195)
> Service Pack 4 machine using MSVC 6.0 (SP5), I ran "testall.exe -v" to see
> how things went.  Alas, some things didn't look kosher.  I cut/paste the 
> output
> for your review.  Please note the following errors:
>  
> testdir             : |Line 236: remove temp dir (720032): The process 
> cannot access the file because it is being used by another process.
> FAILED 1 of 13

Likely because the test tries to remove a directory in use, which is fatal
on some platforms and completely valid in others.  A test bug.

> testfile            : /Line 697: expected <8194>, but saw <8192>
> /Line 761: read should flush buffered write and fail
> FAILED 2 of 28

Consistent with extra \n -> \r\n decoding, but I don't know for a fact
that's the case, or if something else is going on here.

> testlfs             : |Line 32: Large Files tests require Sparse file 
> support

Sure we have sparse files, but APR never accounted for them.  It's a bad
test that relies on platform knowledge.

> testpipe            : |Line 69: expected <0>, but saw <22>
> /Line 91: expected <0>, but saw <22>

testpipe tries to use unix blocking i/o which win32 doesn't support for
it's pipes.  Win32 apr.h reflects this, so the test is buggy.

> ^C
>  
> Notice the ^C after the testpipe testcase.  I had to break out of that
> because the program was hung.  I also ran the testall.exe program from 
> within
> the IDE, and still testpipe hung, so I also cut/paste the callstack after
> breaking and include it here for your review:
>  
> NTDLL! 77f88a87()
> read_with_timeout(apr_file_t * 0x00c919a8, void * 0x00c91998, unsigned 
> int 14, unsigned int * 0x0012fee0) line 74 + 32 bytes

I've started to struggle with this and run out of time to investigate each
time, your posted research and any further research you can invest are both
greatfully appreciated by this list.

Yours,

Bill

Mime
View raw message