Re: Web headers

Re: Web headers

[Andrea 'Fyre Wyzard' Bocci]

At 14.24 17/01/2002 (GMT +0100), Gerald Pfeifer wrote:

>If you have any further questions, do not hesitate to ask. When working on
>these cleanup patches, I'd start with the main directory before attacking
>subdirectory of htdocs, and first submit patches for only a few pages at
>a time.
>
>Gerald

Actually, I was thinking to commit jus a patch when I have finished.
I'll submit one per directory, if that's better.

I'm putting the pages online at 
<http://studenti.fisica.unifi.it/~fwyzard/gcc>.
(The whole htdocs tree is there, even the not-yet-patched pages).
Before issuing a patch, I'd like to have some feedback from people using 
various browsers - for now, I'm only using IE 6.

fwyzard

P.S.
I'm having problems with SMTP, so I hope not to have sent this twice.

Re: Web headers

[Andrea 'Fyre Wyzard' Bocci]

At 14.24 17/01/2002 (GMT +0100), Gerald Pfeifer wrote:

>If you have any further questions, do not hesitate to ask. When working on
>these cleanup patches, I'd start with the main directory before attacking
>subdirectory of htdocs, and first submit patches for only a few pages at
>a time.
>
>Gerald

Actually, I was thinking to commit jus a patch when I have finished.
I'll submit one per directory, if that's better.

I'm putting the pages online at 
<http://studenti.fisica.unifi.it/~fwyzard/gcc>.
(The whole htdocs tree is there, even the not-yet-patched pages).
Before issuing a patch, I'd like to have some feedback from people using 
various browsers - for now, I'm only using IE 6.

fwyzard

Re: Web headers

[Gerald Pfeifer]

On Thu, 17 Jan 2002, Andrea 'Fyre Wyzard' Bocci wrote:
> I volunteer to try to make all the pages "XHMTL 1.0 Transitional" compliant.
> At least, all the pages from wwwdocs/htdocs.

You are most welcome!

> Am I right in guessing I must run wwwdocs/bin/preprocess to get them the
> way they are on gcc.gnu.org/ ?

Yes, just set SOURCETREE and DESTTREE in your environment (or the script
itself) and run the wwwdocs/bin/preprocess. All you need is MetaHTML in
your path.

If you have any further questions, do not hesitate to ask. When working on
these cleanup patches, I'd start with the main directory before attacking
subdirectory of htdocs, and first submit patches for only a few pages at
a time.

Gerald
-- 
Gerald "Jerry" pfeifer@dbai.tuwien.ac.at http://www.dbai.tuwien.ac.at/~pfeifer/

Re: Web headers

[Andrea 'Fyre Wyzard' Bocci]

I volunteer to try to make all the pages "XHMTL 1.0 Transitional" compliant.
At least, all the pages from wwwdocs/htdocs.

Am I right in guessing I must run wwwdocs/bin/preprocess to get them the 
way they are on gcc.gnu.org/ ?

fwyzard

Re: Web headers

[Gerald Pfeifer]

On Mon, 14 Jan 2002, Joe Buck wrote:
> Gerald has done some validation, but to really have correct XHTML the
> DOCTYPE should be present.

Right.  And while I have been trying to push XHTML for modifications and
new pages (and cleaned up old stuff whenever I've been doing work "around
the corner"), we still have lots of old pages.

Verifying and updating all of these would be quite a lot of work, as far
as I can see.

On Tue, 15 Jan 2002, Zack Weinberg wrote:
> Observationally speaking, the _only_ thing on most of the webpages
> that isn't strict XHTML (after cleanup) is the <h1 align="center">
> tag.  Convert that to <h1 style="text-align: center"> or whatever the
> CSS is, and the effect will be that some browsers don't center the H1
> content.  No great loss.

Actually, now that we have redefined <h1> in style.mhtml, all occurrences
of <h1 align="center"> everywhere else could/should be changed to plain
<h1>, as you had suggested.  (Preapproved, by the way <hint><hint>!)

> I've been playing with CSS for more serious layout (replacing tables
> with it) on my own webpages.

Do you have an URL handy?

On Tue, 15 Jan 2002, Mark Mitchell wrote:
>> I would recommend only going for "XHMTL 1.0 Transitional".
> I agree.

Definitely. XHTML 1.0 Strict is really a bit too ambitious for us, for the
foreseeable future at least. ;-)

> For other reasons, we've done a fair amount of playing around with
> various browsers, and XHTML 1.0 Transitional seems to work pretty well
> with moderately recent versions of IE, Netscape, and Mozilla.  I
> have not, however, tried Lynx to make sure that works OK, which is
> probably important.

I have also tried it with Lynx, and it seems to work pretty well.

Gerald
-- 
Gerald "Jerry" pfeifer@dbai.tuwien.ac.at http://www.dbai.tuwien.ac.at/~pfeifer/

Re: Web headers

[Mark Mitchell]

> This, on the other hand, is more likely to cause problems: if all of the
> style is in the style sheet, that's more likely to cause problems for
> Netscape 4.x.  I would recommend only going for "XHMTL 1.0 Transitional".

I agree.

For other reasons, we've done a fair amount of playing around with
various browsers, and XHTML 1.0 Transitional seems to work pretty well
with moderately recent versions of IE, Netscape, and Mozilla.  I
have not, however, tried Lynx to make sure that works OK, which is
probably important.

--
Mark Mitchell                   mark@codesourcery.com
CodeSourcery, LLC               http://www.codesourcery.com

Re: Web headers

[Zack Weinberg]

On Mon, Jan 14, 2002 at 03:49:23PM -0800, Joe Buck wrote:
> > Browsers seem to be happy with XHTML <tag />
> > notation in a document marked as plain HTML, but they may not accept
> > the other way around.
> 
> The space between the tag and the / is what makes the backward
> compatibility work: the older browsers just ignore the / as an
> unknown pseudo-attribute.  XML would allow the writing of <tag/>.

Yah.  My point is that switching style.mhtml to an XHTML DTD might
make some browsers object to the pre-XHTML markup which still appears
on some of the webpages.

> > [myself, I'd be quite happy to see us go whole hog for strict XHTML
> > and style sheeting.]
> 
> This, on the other hand, is more likely to cause problems: if all of the
> style is in the style sheet, that's more likely to cause problems for
> Netscape 4.x.  I would recommend only going for "XHMTL 1.0 Transitional".

Observationally speaking, the _only_ thing on most of the webpages
that isn't strict XHTML (after cleanup) is the <h1 align="center">
tag.  Convert that to <h1 style="text-align: center"> or whatever the
CSS is, and the effect will be that some browsers don't center the H1
content.  No great loss.

I've been playing with CSS for more serious layout (replacing tables
with it) on my own webpages.  I don't think the audience is ready for
us to do that to GCC's webpages - but the only page that has tabled
layout is index.html; there are other uses of <table> in our webpages
but they're all real tables.  Okay, search.html is arguable.

zw

Re: Web headers

[Joe Buck]

Zack writes:

> No one's done validation on all the existing web pages to make sure
> they are correct XHTML.

Gerald has done some validation, but to really have correct XHTML the
DOCTYPE should be present.

> Browsers seem to be happy with XHTML <tag />
> notation in a document marked as plain HTML, but they may not accept
> the other way around.

The space between the tag and the / is what makes the backward
compatibility work: the older browsers just ignore the / as an
unknown pseudo-attribute.  XML would allow the writing of <tag/>.

As best I can tell, browsers have always completely ignored headers
they don't understand, and this was a widely understood design principle
of the web going back to pre-Mosaic days.

> Also, I believe there were concerns about browsers giving up
> completely when faced with a DOCTYPE they didn't understand, or the
> required <?xml ...?> header.

I would be amazed if such a browser existed.  Browsers have never
worked that way; they keep lumbering on, trying to do something
sensible in the presence of the most bletcherous error-filled "HTML";
they do so by ignoring what they don't understand, but still processing
any data between <mystery-directive> and </mystery-directive>.

> [myself, I'd be quite happy to see us go whole hog for strict XHTML
> and style sheeting.]

This, on the other hand, is more likely to cause problems: if all of the
style is in the style sheet, that's more likely to cause problems for
Netscape 4.x.  I would recommend only going for "XHMTL 1.0 Transitional".

> zw
> 

Re: Web headers

[Zack Weinberg]

On Mon, Jan 14, 2002 at 12:12:40PM +0000, Joseph S. Myers wrote:
> On Sun, 13 Jan 2002, Russ Allbery wrote:
> 
> > It makes the page valid XHTML, which I expect is more important than
> > strict HTML 4.0 compliance since that's the standard moving forward.
> 
> The bug would be that, since we're moving to XHTML, the DOCTYPE header in
> style.mhtml is an HTML 4.0 one rather than an XHTML one.  (contribute.html 
> includes the appropriate XHTML header.)

No one's done validation on all the existing web pages to make sure
they are correct XHTML.  Browsers seem to be happy with XHTML <tag />
notation in a document marked as plain HTML, but they may not accept
the other way around.

Also, I believe there were concerns about browsers giving up
completely when faced with a DOCTYPE they didn't understand, or the
required <?xml ...?> header.

[myself, I'd be quite happy to see us go whole hog for strict XHTML
and style sheeting.]

zw

Re: Web headers

[Joseph S. Myers]

On Sun, 13 Jan 2002, Russ Allbery wrote:

> It makes the page valid XHTML, which I expect is more important than
> strict HTML 4.0 compliance since that's the standard moving forward.

The bug would be that, since we're moving to XHTML, the DOCTYPE header in
style.mhtml is an HTML 4.0 one rather than an XHTML one.  (contribute.html 
includes the appropriate XHTML header.)

-- 
Joseph S. Myers
jsm28@cam.ac.uk

Re: Web headers

[Russ Allbery]

Andrea 'Fyre Wyzard' Bocci <fwyzard@inwind.it> writes:

> After soem more browsing, I've seen that this "error" is quite common in
> the gcc pages, ie. ending
>    <meta ...
> with
>    ... />
> instead of
>    ... >
> I'm not an HTML serious user - is there any particular reason for it ?

It makes the page valid XHTML, which I expect is more important than
strict HTML 4.0 compliance since that's the standard moving forward.

-- 
Russ Allbery (rra@stanford.edu)             <http://www.eyrie.org/~eagle/>

Web headers

[Andrea 'Fyre Wyzard' Bocci]

As I said before, there are some minor bugs with the headers that are 
automatically inserted in the web pages.
I think they come from wwwdocs/htdocs/style.mhtml.
So, here's a patch that should fix that.

After soem more browsing, I've seen that this "error" is quite common in 
the gcc pages, ie. ending
   <meta ...
with
   ... />
instead of
   ... >
I'm not an HTML serious user - is there any particular reason for it ?

fwyzard


Index: style.mhtml
===================================================================
RCS file: /cvsroot/gcc/wwwdocs/htdocs/style.mhtml,v
retrieving revision 1.19
diff -c -3 -p -r1.19 style.mhtml
*** style.mhtml 7 Jan 2002 14:01:26 -0000       1.19
--- style.mhtml 14 Jan 2002 01:36:34 -0000
***************
*** 21,28 ****
   <define-container HEAD>
   <verbatim>
   <head>
! <meta http-equiv="Content-Type" content="text/html; charset=iso-8859-1" />
! <link rev="made" href="mailto:gcc@gcc.gnu.org" />
   </verbatim>
   %body
   <verbatim>
--- 21,28 ----
   <define-container HEAD>
   <verbatim>
   <head>
! <meta http-equiv="Content-Type" content="text/html; charset=iso-8859-1" >
! <link rev="made" href="mailto:gcc@gcc.gnu.org" >
   </verbatim>
   %body
   <verbatim>