Dmitry Nezhevenko | 1 Jun 09:15 2012
Picon

Re: A question about multiple changes

On Thu, May 31, 2012 at 09:49:05AM -0700, Mike Winter wrote:
> What can I provide from a home-grown mail-parser that indicates multiple owners for a change? 
> 
> wrong---->Change=(who = ['bob','alice']....) 
> 
> breaks buildbot AFAIK with some exception in waterfall or html 

Probably no. Btw why you need this? You want to have other users (like
reviewer, etc) in blamelist too?

I think that the only thing that really matter is email notification. And
CC other users related to commit may be good. Some times ago we were
adding such information to commit message. And once build is done, it's
possible to parse it in buildbot somewhere in custom email format
function and send email to required users (probably add recipients
requires very small patch to buildbot, I can take a look if you need).

Regards
--

-- 
WBR, Dmitry
------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and 
threat landscape has changed and how IT managers can respond. Discussions 
will include endpoint security, mobile security and the latest in malware 
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
_______________________________________________
(Continue reading)

Georges Racinet | 2 Jun 21:09 2012
Picon

Re: Status of hg polling


On May 29, 2012, at 8:07 PM, Moore, Mathew L wrote:
>> After a quick read of GitPoller and its tests, that doesn't seem very
>> hard to adapt to Mercurial. What it relies on is fetching the changes
>> by calling the git executable in a subprocess.
>>
> <snip>
>>
>> Conversely, a generic HgPoller would make my setups more standard.  
>> They
>> are currently heavily based on custom mirroring to gain control of  
>> the
>> hooks while letting the reference repositories untouched.
>>
>
>
> Interestingly, I just today signed up for the mailing list to also  
> ask about the status of Mercurial polling.  Searches online only  
> pulled up links like the pushlog ones mentioned.

That is a funny coincidence, indeed :-)

>
> I take it from the discussion that there is no direct solution  
> available, but is there a "best practice" or suggestion for handling  
> the case where one does not have access to the Mercurial  
> installation where the repo is hosted (i.e., no hgbuildbot.hook)?

What I've been doing so far is setiing the hook in a local clone and  
perform pulls in a cron job. Indeed, these hooks react in the same way  
(Continue reading)

Dustin J. Mitchell | 2 Jun 22:05 2012
Picon

Re: db.schedulers.classifyChanges relies on integrity check

Harry sent along the traceback and logs, and I had a look.

I think what's happening here is that the same change is classified by
the same scheduler twice, as suggested as a possibility in

master/buildbot/schedulers/basic.py
148         # NOTE: this may double-call gotChange for changes that
arrive just as
149         # the scheduler starts up.  In practice, this doesn't hurt anything.

It looks like postgres is aborting the transaction after the first
failed query, which is annoying but known:

dustin=# begin;
BEGIN
dustin=# insert into foo values (1);
INSERT 0 1
dustin=# insert into foo values (1);
ERROR:  duplicate key value violates unique constraint "foo_d_key"
dustin=# update foo set d=2;
ERROR:  current transaction is aborted, commands ignored until end of
transaction block

I suspect that this can be fixed by ending the transaction and
starting a new one.  But, we should be revealing this problem with
some tests, first.

Moved to http://trac.buildbot.net/ticket/2308

Dustin
(Continue reading)

Patrick Marty | 7 Jun 15:49 2012
Picon

Buildbot, mercurial and incoming hook

Hi everyone,

I am using buildbot with mercurial and I want to know why it is not
possible to use
the hgbuildbot hook with a 'incoming' hook, as explained here :

http://buildbot.net/buildbot/docs/current/manual/cfg-changesources.html#mercurial-hook

"The hgbuildbot.hook function will only work with the changegroup hook."

Is it possible to overcome this limitation ?
Is there a buildbot hook available for the 'incoming' hook type ?

Yours
Patrick

------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and 
threat landscape has changed and how IT managers can respond. Discussions 
will include endpoint security, mobile security and the latest in malware 
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
Derek Graham (ML | 7 Jun 18:22 2012
Picon

Spare checkout leads to deleting build directory and another checkout

I'm doing a sparse check-out using the following bit of code:

svn_min_steps = [source.SVN(svnurl = svn_trunk,
                            mode = "update", depth = "immediates"),
                 source.SVN(svnurl = "%s/a" % svn_trunk,
                            mode = "update", depth = "files", workdir = "build/a"),
                 source.SVN(svnurl = "%s/b/c" % svn_trunk,
                            mode = "update", depth = "infinity", workdir = "build/b/c")
                 ]
...
factory.addSteps(svn_min_steps)
... etc

But what I get is that for every build the second and third steps removes the path specifed by 'workdir' then do the whole check-out again rather than update :(

I've tried using baseURL and set defaultBranch to an empty string as I seen that in another post, but that didn't work either.

Removing the first source step didn't help either.

What is the proper method to perform an sparse update?

Many thanks,

d

------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and 
threat landscape has changed and how IT managers can respond. Discussions 
will include endpoint security, mobile security and the latest in malware 
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
_______________________________________________
Buildbot-devel mailing list
Buildbot-devel@...
https://lists.sourceforge.net/lists/listinfo/buildbot-devel
Georges Racinet | 7 Jun 18:44 2012
Picon

Codebases and ForceScheduler

Hi,

I'm playing a bit with the new system for multiple repositories.

On one hand, it seems that specifying codebases in the scheduler is  
mandatory if there are steps using codebases. Otherwise, one gets the  
following error:
           No sourcestamp found in build for codebase 'test-codebase'
Everything works fine for normal operation, but how is that supposed  
to work for ForceScheduler ?

On the other hand, the ForceScheduler does not have the 'codebases'  
keyword argument, which leads systematically to the error above.

So, am I doing something wrong with this new system, or should I file  
a bug for the ForceScheduler ?

Regards,

------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and 
threat landscape has changed and how IT managers can respond. Discussions 
will include endpoint security, mobile security and the latest in malware 
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
Fredric Dorothy | 7 Jun 22:25 2012
Picon

Blank properties

I have a question about property defaults.  From the documentation I
found that I can do something like this:

f.addStep(ShellCommand(command=[ 'foo', Property('bar', default='') ]))

What I would like to happen is for command to be ['foo'] when the
property bar is not defined.  However, command turns out to be ['foo',
''], and that causes the script 'foo' to think it has an extra
argument passed to it.

Is there an easy way to remove empty strings from the command list
during the build?  Wouldn't it make sense to have the command list
automatically remove empty strings as the default behavior?

Thanks,
Fredric

------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and 
threat landscape has changed and how IT managers can respond. Discussions 
will include endpoint security, mobile security and the latest in malware 
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
Georges Racinet | 8 Jun 01:29 2012
Picon

Re: Blank properties

On 06/07/2012 10:25 PM, Fredric Dorothy wrote:
> I have a question about property defaults.  From the documentation I
> found that I can do something like this:
>
> f.addStep(ShellCommand(command=[ 'foo', Property('bar', default='') ]))
>
> What I would like to happen is for command to be ['foo'] when the
> property bar is not defined.  However, command turns out to be ['foo',
> ''], and that causes the script 'foo' to think it has an extra
> argument passed to it.

I've had the same problem, too. Fortunately, in my case, I could set an 
environment variable instead of a passing an argument, and that had the 
wished behaviour if empty.

Worse : how to issue ['foo', '--option=value_from_prop'] or just ['foo'] 
? And about ['foo', '-o', 'value_from_prop'] ?
I guess for these cases an auxiliary script cannot be avoided without 
utter complication.

> Is there an easy way to remove empty strings from the command list
> during the build?  Wouldn't it make sense to have the command list
> automatically remove empty strings as the default behavior?

I don't know any program that gives a special meaning to empty 
arguments. If some people consider this to be an issue, we could use 
None, or a special marker instead of '', to the same effect.

Regards,

--

-- 
Georges Racinet
Anybox SAS, http://anybox.fr

------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and 
threat landscape has changed and how IT managers can respond. Discussions 
will include endpoint security, mobile security and the latest in malware 
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
Dustin J. Mitchell | 8 Jun 08:38 2012
Picon

Re: Blank properties

On Fri, Jun 8, 2012 at 1:29 AM, Georges Racinet <gracinet@...> wrote:
> I don't know any program that gives a special meaning to empty
> arguments. If some people consider this to be an issue, we could use
> None, or a special marker instead of '', to the same effect.

Actually, lots of programs give special meaning to empty arguments.

This is already supported -- use [] for a default value.  Buildbot
"flattens" command lists before executing them.

I think this was added in 0.8.6, so ymmv for older versions :)

Dustin

------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and 
threat landscape has changed and how IT managers can respond. Discussions 
will include endpoint security, mobile security and the latest in malware 
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
Harry Borkhuis | 8 Jun 09:47 2012
Picon

Re: Codebases and ForceScheduler

Georges,

Indeed, when you use codebases in your steps then the scheduler must know these codebases as the step will
ask for sourcestamp information for that specific codebase.
Did you find all the information you need from the documentation?

The Forced Scheduler does not support codebases yet. 
There has been done some effort but it takes more than only the scheduler as the GUI has to be changed to supply
information per codebase. 
As far as I know Jared Grubb has done some development for this, but I don't know the status of that.

Harry Borkhuis

-----Original Message-----
From: Georges Racinet [mailto:gracinet@...] 
Sent: donderdag 7 juni 2012 18:45
To: buildbot-devel@...
Subject: [Buildbot-devel] Codebases and ForceScheduler

Hi,

I'm playing a bit with the new system for multiple repositories.

On one hand, it seems that specifying codebases in the scheduler is mandatory if there are steps using
codebases. Otherwise, one gets the following error:
           No sourcestamp found in build for codebase 'test-codebase'
Everything works fine for normal operation, but how is that supposed to work for ForceScheduler ?

On the other hand, the ForceScheduler does not have the 'codebases'  
keyword argument, which leads systematically to the error above.

So, am I doing something wrong with this new system, or should I file a bug for the ForceScheduler ?

Regards,

------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT
managers can respond. Discussions will include endpoint security, mobile security and the latest in
malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
_______________________________________________
Buildbot-devel mailing list
Buildbot-devel@...
https://lists.sourceforge.net/lists/listinfo/buildbot-devel

------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and 
threat landscape has changed and how IT managers can respond. Discussions 
will include endpoint security, mobile security and the latest in malware 
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/

Gmane