labs-labs mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Leo Simons <>
Subject Re: [report]
Date Wed, 14 Mar 2007 11:05:06 GMT
On Mar 14, 2007, at 8:55 AM, William A. Rowe, Jr. wrote:
> Justin Erenkrantz wrote:
>> To be a bit more concrete (names removed to protect the innocent - if
>> they wish to reveal their cloaks, they can), in httpd-land, a number
>> of developers raised the concern that a committer should no longer be
>> allowed to create a sandbox within the HTTP Server PMC because that's
>> now the charter of the Labs.  The original committer was intending to
>> start on a new version of httpd and wanted to try out some ideas and
>> get feedback from dev@httpd - but, these individuals mentioned that
>> they felt that such a proposal ought to be governed by the Labs PMC
>> *not* by the HTTP Server PMC.  (I believe in the end, the sandbox was
>> created within the httpd PMC SVN space.)
> To clarify, said anonymous committer was told a copy of the project,
> modulo their changes was welcome in a sandbox.

Sorry, but it's not clarifying, Bill. Told by whom, told why, and why  
does labs@ care? I also don't understand the "modulo"-ness.

> The particular case they wanted was to mess with one source file not
> in a project tree, to prove a concept, and the committer didn't even
> expect the source code they planned to commit to necessarily work or
> become part of the project, simply to illustrate that something could
> be done.
> All of which most seems to fall into the scope of a labs. experiment.

*shrug*. Maybe, maybe not. It sounds like an experiment with an  
existing codebase, so I would think it would normally, you know, be  
kept close to that codebase. Just because you can do something like

   svn cp -m "Starting some-idea" https://.../trunk https://.../ 
   svn cp ...-trunk ...-some-idea
   cd ...-some-idea
   svn switch https://.../branches/some-idea
   vi some.file
   vi some.other.file
   svn ci -m "WDYT?" some.file some.other.file

which takes a lot less energy/bandwidth/discussion/etc than setting  
up some separate project (that you first need to discuss with a  
separate group of people), results (probably) in better peer review  
due to the commit going to the right mailing list, etc etc etc.

Of course the HTTPD PMC can decide to disallow such things if they  
want, but I certainly think that this kind of toying around can be  
quite healthy and I'd support it if I was on the HTTPD PMC.

> I don't think most committers are confused by the difference between
> sandboxes and labs.

Yup -- Apache Labs is a specific project with a specifc charter and  
specific rules that hosts specific things, whereas "sandbox" is a bit  
of a vague term we use across the ASF to sort-of mean "you can play  
around and mess with some things here".

> And heck, if you want to sandbox an experiment and encounter  
> resistance
> by your project, you SHOULD proceed to prove your experiment at the
> labs and bring back the finished work to your peers.

Actually, you should then feel free to do a whole variety of things,  
and bringing the experiment to the labs is only one of many things  
you could consider. You could blog about it, start a project on SF or  
google code, or decide to give up open source development alltogether  
and go travel around the globe on foot.

I personally would probably proceed to try and prove my experiment by  
writing a long e-mail or two, posting a patch to jira, or something,  
and I would then stamp my virtual feet and be grumpled with my fellow  
developers for not just seeing through my little experiment. I don't  
think I would approach the labs project. Of course, that's all pretty  
hypothetical, since I've never really worked on a project where any  
of my experiments encountered a lot of resistance.


To unsubscribe, e-mail:
For additional commands, e-mail:

View raw message