Re: [Pljava-dev] building on centos 6.8

On 02/22/17 15:01, John R Pierce wrote:
> ...
> [ERROR] /usr/pgsql-9.5/include/server/port/atomics/generic.h: In function
> ‘pg_atomic_add_fetch_u32_impl’:
> [WARNING] /usr/pgsql-9.5/include/server/port/atomics/generic.h:247: warning:
> conversion to ‘uint32’ from ‘int32’ may change the sign of the result
> [ERROR] /usr/pgsql-9.5/include/server/port/atomics/generic.h: In function
> ‘pg_atomic_sub_fetch_u32_impl’:
> [WARNING] /usr/pgsql-9.5/include/server/port/atomics/generic.h:256: warning:
> conversion to ‘uint32’ from ‘int32’ may change the sign of the result
> ...

Hi,

Only time for a quick comment at the moment, but all of the messages
you've pasted here are the aggressive sign-conversion warnings that
gcc spits out (because the Maven nar plugin enables such aggressive
warnings by default). None of those will be the reason your build
broke.

Because you got a FAILED in the summary at the end (it would have
been SUCCESS, if nothing but sign conversion warnings), there WILL
be a message, buried somewhere in that flood that you got, explaining
why the build broke, but it's not in the excerpt you copied here.
If you can get all the messages in a file and search them, you're
looking for one that probably has the word 'fatal'.

Troubleshooting is easier if you run mvn with the -Pwnosign option
to suppress the flood of sign conversion warnings so you can see
stuff that matters. That and other suggestions that may be helpful
are here:

http://tada.github.io/pljava/build/build.html#Troubleshooting_the_build

Hope this helps,
-Chap

... a couple summers ago, I focused for a few weeks on just rearranging
PL/Java code to make warnings go away, but I gave up on the sign-conversion
ones because many of them were consequences of how things are defined
in PostgreSQL's own .h files, which we're powerless to change.
_______________________________________________
Pljava-dev mailing list
Pljava-dev(at)lists(dot)pgfoundry(dot)org
http://lists.pgfoundry.org/mailman/listinfo/pljava-dev