Re: Deprecation of browser XSLT - SVNIndexXSLT

Daniel Sahlberg <[email protected]> Thu, 23 Jul 2026 14:42:31 +0200

Newsgroups	gmane.comp.version-control.subversion.devel
Message-ID	<CAMHy98N12cQZ5uODryaqVfd_jLNeumPGhP9YTWUctTy1VAUrSw@mail.gmail.com>

Den tors 23 juli 2026 kl 13:56 skrev C. Michael Pilato <[email protected]
>:

> On Thu, Jul 23, 2026 at 3:07 AM Daniel Sahlberg <
> [email protected]> wrote:
>
>> Just from skimming through the docs it looks promising. If you can make
>> it work, I'd really love to put that into the FAQ (and probably as a news
>> item as well).
>>
>>
> I still think we should at least entertain the most low-hanging of the
> requested fruit.  I mean, it would probably take less time and effort than
> this conversation has to add some decoration to mod_dav_svn's HTML output
> that would have no effect on the visual representation of the directory
> listing, but provide better "hooks" into the DOM for folks to tap into.
> Consulting the list previously requested, these two are the least
> disruptive in my opinion:
>
>    - class/id on list, class on list items preserving whether item is
>    dir/file/repository
>    - wrap list in div (would allow a flex layout without JS DOM
>    manipulation, easy to insert a sibling to the list using HTMX)
>
> I again suggest that a simple SVNIndexCSS directive which optionally
> injects a stylesheet reference to the <head> of the output would be
> low-hanging enough to allow folks to make the list prettier.  But I'm also
> not particularly keen to add esoteric features nobody has asked for. :-)
>

I agree, except that I'd prefer naming it SVNIndexInclude to allow the end
user to insert whatever they want (<script>, <meta>, <style>, ...).
SVNIndexCSS seems a bit too limited, especially if it would only contain
the name of the CSS file (much like to SVNIndexXSLT) and just output the
URL in a <style> element.

Is it enough to just dump the content of SVNIndexInclude into the HTML
output or do we also need the ability to load the content from a file? That
would make it an order of magnitude more complicated though (and I'm not
even suggesting being able to load the content from a repository URL...).

Let's get 1.15.0 out of the box first and I can try to take a look.

Cheers,
Daniel