Re: [Lynx-dev] lynx beta bug in displaying xhtml5 pages

[Shlomi Fish]

On Mon, Feb 17, 2025 at 11:36 PM Thorsten Glaser <tg@mirbsd.org> wrote:
>
> On Mon, 17 Feb 2025, Shlomi Fish wrote:
>
> >https://www.shlomifish.org/philosophy/philosophy/putting-cards-on-the-table-2019-2020/index.xhtml
>
> If you want a webbrowser to display it, put the spaces before the />
> and all that. The XHTML/1.1 standard explicitly addresses this. But
> you’ve been proving your lack of interest in this time and time again
> on this mailing list, so I won’t even bother replying again until you
> fix your site.
>

Hi Thorsten and everyone,

thanks for your reply.

Summary: I "fix"ed part of my site and converted "/>" to " />" and it
is still broken
in lynx. Moreover, firefox and chromium (and other browsers) handled the pages
there fine (and still do).

You claimed that the beta version of lynx has hidden most of the text of the
problematic XHTML 5 pages on shlomifish.org , because their markup was served
with standalone XML tags that didn't have whitespace before the slash in the
"/>".

I set out to test this hypothesis. Based on src.xhtml (originally a copy of
the page
https://www.shlomifish.org/philosophy/philosophy/putting-cards-on-the-table-2019-2020/
) I generated, using generate.py, 8 versions with all combinations of whitespace
for "/>" . Turns out lynx-beta erratically hides most of the text on
all of them:

https://github.com/shlomif/lynx-xhtml5-test/tree/master/dest-lynx-test

I also wrote a similar script to mutate the markup of the XHTML pages
(those ending
with “.html” or “.xhtml” ) on my site under /philosophy/philosophy/
to put a space before the slash in the "/>" . It is currently invoked
at every "make test"
stage, and its changes are
present in the live pages under
https://www.shlomifish.org/philosophy/philosophy/
(I checked). the beta version of lynx still does not display most of
the text for

~/apps/lynx-beta/bin/lynx
https://www.shlomifish.org/philosophy/philosophy/putting-cards-on-the-table-2019-2020/index.xhtml

( This was done given your request for me to "fix" the pages of my
site and given I
do not wish to risk potentially breaking the rest of the site with such a
crudely-implemented change for now. )

Moreover note that as far as I recall, the recommendation to prefix the trailing
slash with space was intended for compatibility with legacy www
browsers that had not
been adapted to XML or XHTML. If the page is served as
“Content-Type: application/xml+xhtml” , then a compliant user agent should parse
it as XML. Moreover, most of the pages on my site are validated as XHTML5
using several validators, and firefox and chromium-browser usually
handle them well.

(

Stripping the whitespace in question is done by the X/HTML minifier I
use, and I might
opt to disable or undo it somehow if it increases compatibility with
lynx and other browsers.

Nevertheless: https://caniuse.com/xhtml and
https://knowyourmeme.com/memes/what-year-is-it and
adding whitespace still doesn't seem to help here. I think lynx is
doing the wrong thing.

)

Previous discussion on similar issues which were fixed, but similar
problems emerged
again:

https://src.fedoraproject.org/rpms/lynx/pull-request/4

https://lists.nongnu.org/archive/html/lynx-dev/2022-06/msg00016.html

Here is a screen recording of lynx beta on fedora 41:

https://github.com/shlomif/lynx-xhtml5-test/blob/master/termtosvg-lynx-beta-on-fedora41.svg



> bye,
> //mirabilos
> --
> Yay for having to rewrite other people's Bash scripts because bash
> suddenly stopped supporting the bash extensions they make use of
>         -- Tonnerre Lombard in #nosec



-- 
Shlomi Fish https://www.shlomifish.org/

Things will always go wrong, even if they cannot.

Please reply to list if it's a mailing list post - http://shlom.in/reply .