Shreedhar Ganapathy wrote:
>> I updated my workspace and did another build and it worked. The
>> update did
>> get new versions of a few files, is it possible someone fixed the build
>> problem?
> The cvs tip is open to checkins so more files could have been checked
> in. But those were not related to your build errors. Your build somehow
> did not have the correct version of shoal-gms.jar. That's the only
> reason for those errors. The cvs mailing list archive shows when the
> project.properties under glassfish/bootstrap was updated with the new
> version information for the revised shoal jar and that's around midnight
> Friday night.
>> Or is it possible that my checkout caught things in some
>> intermediate state?
> That is most probably the case. But that could have only happened if the
> checkout happened around midnight though.
>> I did the checkout in the afternoon.
> Huh? I dont get it then. I dont see any checkins to fix a build though.
Just another example of the unreliability of our build process?
I did the build just before 7:00pm on Saturday. I can send you the
entire build output if someone wants to try to find the root cause.
(I'm afraid I no longer have the workspace in the same state because
of subsequent updates and builds.)
>> So, what about the other issues I pointed out?...
>
> I agree that filtering the expected/known warnings would allow
> unexpected warnings/issues in build to be caught quickly. But I would
> prefer if each module could explore addressing these warnings the right
> way rather than inventing a way to hide/filter these warnings. Post beta
> this could be part of each module's todos. Another area where
> contributions from community would help.
Maybe I wasn't clear...
I think the warnings should be fixed wherever possible.
What I wanted to filter out was the annoying "chatter" that's part of
a successful build. For example, the "BUILD SUCCESSFUL" messages, the
"cvs server: Updating" messages, and so on. All the completely
predictable messages that don't indicate any failure at all, but are
only useful for knowing what has happened before a failure occurs.
With most or all of those messages gone, it would be easier to visually
scan the output for messages that indicate real errors. I would expect
the warning messages to appear in such output. (I fear that one reason
the warnings might be being ignored is that if you only look at the end
of the output where it says "BUILD SUCCESSFUL", you won't see them.)
Understand?
Note also that there's a bug in the jdbcra build file that causes the
erroneous output I highlighted in my previous message.