Re: Deprecating <tref>

The 0.0000003% number is from all uses of <tref> (including linked
resources, images, etc.) divided by the total number of pages. It is
surprising to me that the spec authors would push to keep this in the spec
knowing it's almost entirely unused and that the last remaining browser
implementation is actively moving to remove it. What a disservice to SVG
authors.

Dirk,

If <tref> were changed to match <use>'s component model, the implementation
issues around <tref> would go away. As it's spec'ed today, <tref> can
reference any node which is why it has been a burden on our implementation.

Philip




On Thu, Jun 27, 2013 at 12:02 AM, Manuel Strehl <svg@manuel-strehl.de>wrote:

> Hi,
>
> partial reason for the low usage is surely <
> https://bugzilla.mozilla.org/show_bug.cgi?id=273171>. Authors on the web
> don't bother with it, since it's not supported in Firefox.
>
> I actively vote for keeping it. There's, for example, one use case, that
> <tref> can handle and <text> not: localization with a remote text server:
>
> <tref xlink:href="
> http://example.com/translate/My%20hovercraft%20is%20full%20of%20eels"/>
>
> Actually, I'm waiting for years for Firefox to fix #273171 so I can
> finally use <tref> on the web.
>
> Cheers,
> Manuel
>
>
> >
> > On Jun 26, 2013, at 8:00 PM, Doug Schepers <schepers@w3.org> wrote:
> >
> >> Hi, folks-
> >>
> >> Philip's concerns seem reasonable, but I also I hear reluctance to
> >> deprecate a feature that's in use, which I share; I've used <tref> quite
> >> a bit myself back when Adobe still supported their plugin, though less
> >> so in browsers, since support is spotty.
> >>
> >> When we look at <tref>, it largely seems like a special case of <use>,
> >> exclusive to text. Since we are revamping <use> as an application of the
> >> component model, it makes sense to me that we would do the same for
> >> <tref>; that way, it would share a mostly unified code base and security
> >> model with <use>, which itself would derive from the component model,
> >> making it easy to maintain.
> >>
> >> Does that seem like a reasonable solution?
> >
> > I think so. If there is content using it and enough implementations
> > supporting it, we should not deprecate it IMO. We will revisit this
> anyway
> > when we want to get to CR. I support using the same component model for
> > <tref> as for <use>.
> >
> > Greetings,
> > Dirk
> >
> >>
> >> Regards-
> >> -Doug
> >>
> >> On 6/26/13 8:38 PM, Rick wrote:
> >>> Dear Working Group:
> >>>
> >>> Thank you for your efforts and hard work in advancing SVG and in
> >>> maintaining a robust and important specification.
> >>>
> >>> I ask you to consider that, while this proposal may not effect most
> >>> content, it definitely will break leading edge SVG applications
> >>> currently deployed in the air traffic industry.
> >>>
> >>> <tref> is useful for shadowing text.
> >>>
> >>> Consider an application that uses a cursor with coordinates following
> >>> over a multicolored display for mapping, image editing or CAD. Having
> >>> to set only one element speeds the process up and improves the look,
> >>> feel and functionality of the application.
> >>>
> >>> /S/hadowed text with <tref> is used in many less critical areas that
> >>> benefit from this convenience.
> >>>
> >>> If this feature is deprecated, it will affect software used in high
> >>> profile engineering mapping displays deployed at major international
> >>> airports on five continents, soon to be six.
> >>>
> >>> It's a good feature.
> >>>
> >>> Cheers!
> >>>
> >>>
> >>> On Wed, Jun 26, 2013 at 5:15 PM, Philip Rogers <pdr@google.com
> >>> <mailto:pdr@google.com>> wrote:
> >>>
> >>> www-svg,
> >>>
> >>> I would like to propose deprecating <tref> from SVG2. I would also
> >>> like to field your opinion on removing it from Blink.
> >>>
> >>> Our numbers show <tref> use in the wild is virtually nonexistant:
> >>> less than 0.0000003% of pages. Furthermore, the supporting code is
> >>> complex and has been a source of many security bugs in Blink and
> >>> WebKit. Of the 24 tref bugs that have ever been filed against
> >>> Chrome, 14 have been stability or security related.
> >>>
> >>> What do you think of slimming up both the spec and implementations
> >>> by removing <tref>?
> >>> https://svgwg.org/svg2-draft/single-page.html#text-TRefElement
> >>>
> >>> Philip
> >>>
> >>>
> >>>
> >>>
> >>> --
> >>> /Rick Graham.
> >>> /
> >>> /Senior Applications Architect, NAVCanada <http://www.navcanada.ca/>.
> >>> /
> >>> /Contributing Author, Scalar Vector Graphics (SVG)
> >>> /
> >>> /grahari@navcanada.ca <mailto:grahari@navcanada.ca>
> >>> /
> >>> /graham.rick@gmail.com <mailto:graham.rick@gmail.com>
> >>> /
> >>>
> >>
> >>
> >>
> >
> >
> >
>