Re: Requiring DOAP instead of MAINTAINERS file

Am Samstag, den 19.01.2008, 01:28 +0100 schrieb Olav Vitters:
> On Fri, Jan 18, 2008 at 07:40:00PM +0100, Mathias Hasselmann wrote:
> > 
> > Am Freitag, den 18.01.2008, 15:57 -0200 schrieb Tristan Van Berkom:
> > > I can see that xml might be a little less beautiful to the eye, and
> > > also
> > > that text files might be a little less beautiful to a machine who has
> > > to parse some custom format that might be subject to change, the
> > > great 
> > > thing about xml is that is very easy to read for a computer and
> > > for a human. 
> > 
> > Yes, that's how XML was designed, and this promise works pretty well for
> > XHTML and many other XML based language - but producing a XML based
> > language thats human and machine friendly needs some attention. The DOAP
> > examples I've seen so far do not show the slightest evidence, that DOAP
> > was designed for humans. Most parts of a DOAP file just is RDF and
> > namespace boilerplate. Who should remember all that crap? Well, of
> > course a template file could be used for all that boilerplate stuff.
> > Well, unfortunately the need for using template files, is a very good
> > sign for poor language design - IMHO.
> You plan on changing that stuff every second or so?!? IMO after the
> standard DOAP files have been generated, it should be ok to be edited.
> I'll probably ask e.g. GHOP to fill in any missing detail anyway.

So you have a solution for existing projects - although I question the
morality of abusing GHOP students for filling out that binary junk.

New projects still would have to deal with editing highly ugly files.

I agree, that DOAP has a purpose, and those who want to use it, shall
use it. Nevertheless DOAP is a highly ugly format, and I see no
rationale in suggesting and even enforcing usage of a highly ugly file

> You say that e.g. MAINTAINERS is easy. Well, my parser script doesn't
> agree.
> I don't see what is hard about:
> shortdesc>Java-based build tool</shortdesc>

Would be nice, if DOAP would be that simple. But in reality DOAP
requires usage of very long XML namespaces, and usages of very strange
attribute names. Yes, I know - they are defined by RDF - but this
doesn't make them less strange.

> vs:
> shortdesc: Java-based build tool
> Both have to be thoroughly checked for syntax, as maintainers will get
> them wrong, often. Yes, XML is harder, that is why there should be
> examples. It isn't that hard. Further, usually it is only a one-time
> exercise.

So you seriously suggest, that switching to a more complicated format
will cause less problems? Sorry, but were are the cameras?

> > > Also agree with Murray that having a sophisticated centralized
> > > database would confuse and complicate things in gnome, it's somehow
> > > reassuring to me that every project has all of its important
> > > information in its separate tarballs and revision history.
> > 
> > Yes, that central database part of Olav's proposal is a part I failed to
> > understand. Motivation seems less annoyance due commit hooks. Just
> > another reason to use some easy to understand text format, instead of
> > DOAP which seems error prone with all that long namespaces and long
> > attribute names.
> Lots of people don't even get the MAINTAINERS syntax right. I don't
> believe anything will work without strict enforcement (the MAINTAINERS
> syntax check is *really* simple). IMO it is like autotools. Almost
> nobody understands it and just copies some existing file. Which works
> just fine.

Mathias Hasselmann <mathias hasselmann gmx de>

Attachment: signature.asc
Description: Dies ist ein digital signierter Nachrichtenteil

[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]