[OSM-dev] Re: [OSM-talk] OSM web forums?

[Tom Carden]

Off the talk list, and copied to dev (not sure if you read it or not).

On 4/4/06, Jim Ley <jim at jibbering.com> wrote:
>
> Sorry, filters themselves are just as optionally supported in IE, so the
> point is equivalent.
>

So you're saying that because it possibly won't work for all IE users,
we shouldn't do it at all?  I don't think that's how it works around
here.  Perhaps it's worth running a script that reports how many
people the fix would actually work for, just to check?  (Bearing in
mind we already require javascript for the site to work at all right
now).

> >IE7 isn't going to make browser compatibility hacks go away, and if we
> >know how to support IE6 users, why shouldn't we?  [1]
>
> Because you can support it without complicating matters, the very fact that
> those of who use IE are getting script errors shows the problem of using
> scripts,

No, it shows the problem of non-IE users fixing IE bugs, that's all. 
If there hadn't been a script error, would you have bothered to tell
us that pngfix isn't a good solution?

I think that alpha transparency is worth hacking for.  If we start
rendering things like notes/pins or routes on the map we'll almost
certainly want it for those too - not just because drop shadows are
"web 2.0", either.

> the onload change the document behaviour harms rendering,

How so?

> it's much
> better for things to degrade gracefully than continually adding more hacks
> and more scripts for more and more situations - kill the script and no-one
> is harmed at all, with the script people are getting errors.
>

We know how to fix the script, so that's not really a fair
representation of the options, is it?  If we kill the script, we can't
use alpha transparency in PNGs - I'm happy to discuss the merits of
that capability, but I'm highly in favour of it.

I'd also be in favour of removing all cosmetic hacks when IE6 has
fewer users than IE7.

> I use IE, the pngfix fails due to the bug, OSM is completey usable other
> than the script errors - the only logical conclusion is remove the script,
> then there can't be any errors from the code. [1]

That's not the only conclusion.  It's one conclusion.  Another
conclusion is to fix the bug.

>
> > We should keep the option open to do that again.
>
> Not really, it would seem likely that any compositing would be done in
> script and pngfix.js which only works on images that exist when the document
> is loaded (which is of course miles after the images appear to the user) and
> not those that are created later.

You can use a similar technique for dynamically added images.  The
compositing was done in a time before tiles (yes really) so the images
were already in the page anyway.

> > Personally
> >I think it's patronising, and I also don't like the fact that it's yet
> >another Google Toolbar bundling product, but never mind.
>
> I assumed it was for money raising reasons, I'm very disappointed if it's
> some sort of ideological reason.
>

Yes, it is for money raising reasons, but I still find it patronising :)

T.


> Cheers,
>
> Jim.
>
> [1] Well actually a background colour would be nice too, but it is still
> usable if not a little odd looking...
>

What element do you want a background colour on?

Tom.