apr-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Sander Temme <san...@temme.net>
Subject Re: [VOTE] Release apr / apr-util 1.2.8 / 0.9.13
Date Thu, 30 Nov 2006 18:51:40 GMT

On Nov 29, 2006, at 2:44 PM, William A. Rowe, Jr. wrote:

>
>
> William A. Rowe, Jr. wrote:
>> Justin Erenkrantz wrote:
>>>> If we roll over to 1.2.9, I'd like to do it no later than  
>>>> tomorrow evening.
>>> As I said, I don't view these failures as anything that should  
>>> hold up 1.2.8.
>>
>> Understood - but Solaris seems to be the biggest thorn in our side  
>> (and their
>> own - heh).  In combination with the BSD issues, I'd really  
>> appreciate if some
>> folks can jump in and triage these failures - if readily fixed  
>> lets bump on,
>> if not, lets plan a 1.2.9.  But I'd really prefer not to roll out  
>> a 1.2.9 this
>> year.
>
> Should have said - I'd really prefer not to roll out 1.2.9 later  
> this year.
> Let's solve it in this release AS 1.2.9 if we can find the quirks.

I'm not sure what the BSD issues are, but it doesn't look like a  
regression: FWIW if you run the individual tests using this slightly  
modified language from the Makefile:

for p in testlockper testshmproducer testshmconsumer testmutexscope  
testall ; do ./$p > /dev/null 2>&1 ; echo "$p result: [$?]" ; done

You get the following result on that same dual amd64 FreeBSD 6.1- 
RELEASE box:

testlockper result: [127]
testshmproducer result: [254]
testshmconsumer result: [254]
testmutexscope result: [0]
testall result: [0]

Again, FWIW because I don't know what these result codes signify.

I just built APR 1.2.7, with exactly the same results both from make  
check and the above. So, it doesn't look like we have regressions.

So what we see here is that testlockper returns 127, which for some  
reason does not cause make to freak out, but the 254 result does. We  
check for 255 in the Makefile and return "$prog failed" when that  
happens, without additional information, but apparently the 254  
causes make to abort without executing the if block. All of this to  
the detriment of subsequent tests.

Running make -i check causes the suite to complete with no errors...  
looks like the testshmproducer and testshmconsumer tests don't  
generate any output for better or for worse.

I'm +1 on releasing (based on no regressions) but we may want to pay  
some attention to the test suite. Perhaps better discussed under  
separate cover.

S.

-- 
sander@temme.net              http://www.temme.net/sander/
PGP FP: 51B4 8727 466A 0BC3 69F4  B7B8 B2BE BC40 1529 24AF


Mime
View raw message