- From: (unknown charset) Alex Rousskov <rousskov@measurement-factory.com>
- Date: Tue, 9 Mar 2004 10:55:05 -0700 (MST)
- To: (unknown charset) Tex Texin <tex@XenCraft.com>
- Cc: (unknown charset) public-evangelist@w3.org
I agree with Tex. Ideally, there needs to be one validator that is as complete as possible for a given environment. For example, online checks could include link checks and HTTP compliance/robustness checks as an option. If that validator can accept plugins to validate new things, the validator development can be more distributed. However, from a validator user point of view, it does not really matter. I would also add that it would be great if that validator can be installed as a local library of some kind, to be integrated with content development and content generation tools. This requirement implies writing Perl and other wrappers and/or providing a local service with a simple socket interface. The output can be a very simple XML output (error ID, location, informal text), I guess. Sophisticated wrappers can be added in some environments to provide nice rendering and additional features such as links to specs. Alex. On Tue, 9 Mar 2004, Tex Texin wrote: > > Karl, > > I guess the gist of what I would be looking for as an author, and as someone > seeking greater W3C standards compliance, would be a "checker" that is not so > compartmentalized (eg only does links) and simply reports all possible errors. > > I understand this is a difficult request, but I do think it confuses people > that they run a validator and yet still undetected and common errors remain. > Running separately a validator, and a link checker, and an embedded css > checker, suggests that after a lot of checking, other categories of errors > still exist and it seems to dilute the value of the effort. > > The solution could be as simple as define a common program interface that > allows people to integrate checking tools and have one command that verifies a > page using an extensible list of tools, or perhaps verifies an entire web site. > Others could then write additional checkers that share the interface (eg i18n, > wai, or other checkers). > It would also be easier to integrate checking with authoring tools. (A menu > item could launch a thorough check.) > > As for your question- > a) list all requirements- my understanding is many of the needed checks are on > todo lists... > I think if a start was made on the list of additional checks people would like > to have, plenty of input would be offered. ;-) > > b) example of what a validator should work with- I am not sure I understand the > question. > Perhaps the description above captures the sense of what I would like. > My goal is really not much more than than to identify common errors in pages > with a single tool and not be in the business of collecting multiple tools to > perform different aspects of what might be considered a single task. > > c) a list for discussion would be good. > I would like to see 2 types of discussion. Perhaps it should be separate lists. > 1) list for more thorough and integrated checking tool, as discussed. > > 2) a place for people to discuss obstacles to upgrading to more recent > standards or to being fully compliant, and potential workarounds or solutions. > This might help us understand how to eliminate the (perhaps perceived) > obstacles. > For example, I have heard the arguments for why authors shouldn't specify a new > window vs reusing an existing window. > Nevertheless, I prefer to make the choice in my pages and therefore it is an > obstacle to moving from transitional to strict dtd. I'd like to discuss > solutions so I can be strict without giving up something I see as important. > > Note, I am not trying to argue this particular point now, on this list. I am > giving an example of issues that a discussion list might be useful to develop > ideas to move people forward more quickly. > > hth > tex > > > Karl Dubost wrote: > > > > Le 08 mars 2004, à 14:59, Tex Texin a écrit : > > > It could be the philosophy behind the validator has changed, but in > > > Nov. 2002, > > > I reported that the validator did not check for problems such as links > > > to > > > fragments that do not exist. (e.g. <a href="#tex"> with no > > > corresponding id=tex > > > or name=tex in the document). > > > At the time it was stated that it wasn't a priority for the validator > > > to find > > > problems such as this, unless it was covered by the DTD. i.e. strictly > > > speaking > > > it wasn't a validation problem. > > > > I think "link checker" verifies this type of mistakes: internal links > > of a document > > http://validator.w3.org/checklink > > > > Would you be interested to > > 1. make the list of all requirements MUST, MAY, SHOULD, etc included > > in the HTML Spec. (MUST are already done, see my previous message.) > > 2. give an example on what do you think a validator should work with > > such a list. Implementation requirements for the validator to be able > > to detect or warn users. > > > > It can be a collective effort of the mailing list. Though I'm not sure > > of the right forum for that? public-qa-dev? www-qa? or here? > > > > Olivier, what do you think about it? > > > > -- > > Karl Dubost - http://www.w3.org/People/karl/ > > W3C Conformance Manager > > *** Be Strict To Be Cool *** > > > > ------------------------------------------------------------------------------- > > Name: PGP.sig > > PGP.sig Type: application/pgp-signature > > Encoding: 7bit > > Description: =?ISO-8859-1?Q?Ceci_est_une_signature_=E9lectronique_PGP?= > > -- > ------------------------------------------------------------- > Tex Texin cell: +1 781 789 1898 mailto:Tex at XenCraft.com > Xen Master XenCraft http://www.XenCraft.com > Making e-Business Work Around the World > ------------------------------------------------------------- > >
Received on Tuesday, 9 March 2004 12:55:06 UTC