gump-general mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Stefan Bodewig <bode...@apache.org>
Subject Re: Stefan's comments on Gumpy ( was Re: Sanity Check & Gump Community [please read/respond] )
Date Fri, 26 Sep 2003 15:02:27 GMT
On Fri, 26 Sep 2003, Adam R. B. Jack <ajack@trysybase.com> wrote:

> A +0 at best,

which means "yes, but I won't/can't help".

There have quite been a few people on this list who apreciated a
Python rewrite a few months ago and I've never been among them.  Just
wait for the others to chime in. 8-)

> The Python Gump code is pretty simple, but still takes some time. I
> can understand that this is a barrier to entry for a lot of Java
> programers.

Just to be clear, Java is by far not the first language I've learned
(Sinclair's dialect of Basic IIRC ;-) and it certainly won't be the
last.  I don't need Python for any other stuff I work on (neither for
fun nor for money) and current Gump matches all my needs.  I simply
don't have any reason to learn Python - and even less time.

> As it stands it does the core

The single most important feature for me is that it can run at night
from cron with nobody around - and I love the timeout feature at least
available on Linux that can kill hanging Axis or Cactus or
Tomcat-Connector builds.

I'm pretty much a no-GUI guy, so a nifty GUI won't buy me.

> 1) It doesn't process outputs (copying jars somewhere, publishing
> javadocs, publishing junit).

Yes, this has become an important side effect.

> 2) It doesn't process projects on workspaces only on modules.

I have two extra <module>s in my workspace and a project override for
the "gump" project itself (to add some deliver tags).  It is important
to me but not vital.

> 3) It doesn't handle depends on ant tasks (keep meanign to ask you
> Stefan, 'cos I've seen you use it, what is that?)

I have no idea what you are talking about, sorry.

Maybe

<ant>
  <depend project="..." .../>
</ant>

Take a look here <http://jakarta.apache.org/gump/ant.html#depend>.

>> > *Python Gump using Forrest
>>
>> I don't care at all, honestly.
> 
> Do you run Gump personally/locally,

Yes.  Via cron every night.  And sometimes manually when I want to
test a bigger change to Ant.

> or do you run local custom Gumps?

Basically the default profile plus two extra modules.

> In short -- do you run it remotely and have it communicate to you
> via HTML(/RSS)?

It communicates to me via email.  <nag to="my-address/> in my
workspace.  I only seldomly look at the generated HTML output, I have
never looked at the RSS output.  Thus I really can't be bothered how
the output gets generated.

> Do you want HTML output

Yes.

> (whether it is forrest or XYZ that helps produce it?)

Exactly 8-) 

I don't care how it is produced.

>> > *Gump Statistics (as a tool for automatically communicating
>> > reuse/reliability/robustness, etc.)
>>
>> Depends on the kind of statistics.  I'm really concerned about
>> reliability/robustness values computed from Gump failures.
> 
> Ah, really really? Cool, do I have something to entice you?

There may be a language problem here, I'm not sure.  "concerned" was
supposed have a negative connotation when I wrote it, sorry.

> I'd like statistics to show "responsiveness" (to problems).

This is a type of statistics I'd be most afraid of.

Build of project A fails because they've added a new dependency and
failed to update their descriptor.  I see it first because my
nightlies start a few hours before Sam's.  Do I go ahead and fix the
failure or do I keep away from fixing it to not dilute the statistics?
Do I keep away from fixing it for weeks to show how unresponsive a
project team is?

I wouldn't appreciate any automated finger-pointing either.

> BTW: What bugs me is if product X changes it's API (or fails to run
> unit tests) and hence breaks innocent product Y, then Y gets the bad
> mark. If X is then stagnant/unresponsive, Y gets slammed. I don't
> see a way that Gump can detect it was X that broke Y, and not Z. Now
> I realize that Y will eventually move off X,

Very unlikely, even for projects that consider themselves "friends of
Gump".  When has Gump tried to build Cocoon or Forrest the last time?
Are you aware that almost all krysalis-* projects fail every night
"because Centipede has not been not initialized"? 8-)

>> > *Distributed Gump Trees (also using Ruper to download upstream
>> > Gump outputs)
> 
> If I could automatically (via Ruper) download last night's Gump'd
> outputs from Jakarta, then I'd be "in pretty good shape"

OK, I see.  Yes, having a tool that spiders
http://gump.covalent.net/jars/latest/ and put the jars in place for
your installed modules would help.

> Ok, so the flip side. What say folks did this, and the main gump
> became a 15+ hour build.

We've been close to that already.  When sourceforge's CVS was really
really slow, I mean, even slower than right now.

> Would you stop running Gump, and use rsynch/Ruper/other to get jars
> from public?

No.

> Would this be bad for you?

Extremely.

> (I guess I don't know what do you use Gump for.

A regression test environment for Ant - mostly.  And as such I have to
build everything.  I understand that I'm in an almost unique position
here.

Stefan

Mime
View raw message