db-derby-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Rick Hillegas <rick.hille...@oracle.com>
Subject Re: Question on how to debug two instances of derby off the trunk
Date Tue, 22 Apr 2014 21:34:20 GMT
On 4/22/14 12:11 PM, Bergquist, Brett wrote:
> So it seems that unless a release build it built, then the scripts are not in the proper
place.  I suppose a snapshot build will be similar to a release build and I have built a snapshot
build before although I do get confused about "sane" and "insane" as the snapshot build complains
when I have no "insane".  Maybe some pointers on "sane" and "insane" and their relationship
to builds would be useful.
Hi Brett,

I haven't tried the snapshot target by itself in a long time. However, 
the main release process does call snapshot eventually. It's likely that 
standalone snapshot is broken. Certainly the instruction here are 
broken: http://wiki.apache.org/db-derby/SnapshotsAndQuickBuilds. Those 
instruction tell you to invoke the prepareforrelease target. But I think 
that target will fail if you don't have commit privilege. You may be 
able to get it to work by setting debug.release.build=true.

Alternatively, you could try generating a full release (here again you 
will need to set debug.release.build=true). It's a heavyweight process 
because it builds the docs as well the code. And it requires you to give 
a passphrase for a gpg key in order to sign the artifacts.

The terms sane and insane are very confusing. A sane build is one with 
debug information and an insane build is one without debug information.

It's unclear to me why you need something as heavyweight as a snapshot 
or a full release if you just need to test your changes. Maybe some 
simpler approach would be more productive.

Hope this helps,
> Appreciate the time you take to respond Rick!
> Brett
> -----Original Message-----
> From: Rick Hillegas [mailto:rick.hillegas@oracle.com]
> Sent: Tuesday, April 22, 2014 11:30 AM
> To: derby-dev@db.apache.org
> Cc: Bergquist, Brett
> Subject: Re: Question on how to debug two instances of derby off the trunk
> On 4/22/14 7:31 AM, Bergquist, Brett wrote:
>> I am starting on a project to add the capability to the derby
>> replication to be able to acquire the database to replicate
>> automatically by the slave and I would like to be able to run two
>> instances of the network server so that I can debug the protocol
>> between the two when the slave is going to acquire a database from the
>> master.
>> I guess my question is relating to what to build and where to run it
>> from the source. I have checked out the trunk of derby and am able to
>> build and ran all of the tests successfully. I would like to run the
>> "startNetworkServer" script to run the built binaries but I don't know
>> where the built "startNetworkServer" script is. There is the "bin"
>> directory right at the root of the checkout but that seems to only
>> have the Windows scripts. There is the "generated/bin" directory but
>> the "generated" directory does not seem to have everything else needed.
>> So what is the proper build targets that I need to build and where
>> would the script that I need be located once that is done?
> Hi Brett,
> I think it works this way:
> 1) The source for the Unix versions of the scripts lives in bin/templates
> 2) The source versions are transmogrified by the makebinscript target in the top level
build.xml. The transmogrified versions are dropped into generated/bin
> 3) When we build a release, the targets in tools/release/build.xml build a bin directory
which includes the Windows scripts from bin and the Unix scripts from generated/bin.
> Hope this helps,
> -Rick
>> If someone can give me pointers on what they would do to run both a
>> derby master and derby slave under a debugger that would be really
>> useful.

View raw message