Re: [gentoo-dev] GLEP ??: Critical News Reporting

[Brian Harring <ferringb@gentoo.org>]

[-- Attachment #1: Type: text/plain, Size: 2295 bytes --]

On Tue, Nov 01, 2005 at 10:16:35PM +0000, Ciaran McCreesh wrote:
> On Tue, 1 Nov 2005 22:57:13 +0100 Jan Kundrát <jkt@gentoo.org> wrote:
> | On Tuesday 01 of November 2005 19:25 Ciaran McCreesh wrote:
> | > On Tue, 1 Nov 2005 17:22:29 +0100 Jan Kundrát <jkt@gentoo.org>
> | > wrote:
> | > | What's wrong with XML format similar to the one that is used for
> | > | our GLSAs?
> | >
> | > 1. Portage does not handle XML. Portage will not handle XML in the
> | > near future.
> | 
> | How will it handle GLSAs then? [1]
> 
> gentoolkit != portage.
Correct.  Course, also incorrect.

We already have a module for parsing metadata.xml in use in the 
experimental 2.1 branch (which can be backported to 2.0 if anyone 
wants it and does the work).

Python comes bundled with xml as long as you don't have the build flag 
enabled.



> | > 5. XML is merely adding another problem to the one we have already.
> | 
> | Could you please explain?
> 
> Parsing XML is complicated. Writing XML is complicated. I put together
> a complete working client that can show news items in the plain text
> format proposed -- it took me fifteen minutes to write. I threw
> together a script which can be called after a cron sync that mails news
> items to root in under a minute. There is no way this could be done if
> XML were being used -- any task involving news items would be a major
> chore.

Come on now, this is a bit of bullshit.
http://dev.gentoo.org/~ferringb/portage_metadata.py,
47 lines

Might be a pain in the ass under the frameworks you deal in, but 
something as simple as news items can be done pretty quickly.

> | > I have an eselect module which can read these news files. The whole
> | > thing is about the same size as the DTD would need to be for an
> | > XML-based solution. I have a parser written for the format in
> | > question. The whole thing is smaller than the initialisation code
> | > for an off the shelf XML parser.
> | 
> | Great. Why haven't you just used existing code from `glsa-check`, BTW?
> 
> Because merely figuring out the XML DTD takes longer than it does to
> write an entire client for plain text news items.

Not arguing for, or against, just pointing out certain arguements are 
questionable...
~harring

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]