comprehensively broken for minefield nightlies

[peepo]

install latest nightlies:
http://ftp.mozilla.org/pub/mozilla.org/firefox/nightly/latest-trunk/

visit:
http://www.honte.eu/playGo/games/players-1242215948.xml

this wfm in ff 3.1.10 but not recent nightlies
please consider filing a bug with mozilla if you consider it it not an issue with smil.user.js

regards

~:"

[peepo (j-chetwynd)]

also tested against plain svg in attachment

[codedread (codedread)]

Please describe what is broken about this - it seems to work identically in Firefox 3.6 nightly, Safari 4 and Opera 9.6:

- press the forward arrow to see next moves
- press the backward arrow to undo moves

Providing a minimal test case would help to identify the problem and expedite a fix.

[Lthere (helder-magalhaes)]

I've been testing FakeSmile with recent Namoroka builds (Firefox 3.6b1) and everythings looks fine.

peepo, could you provide more information on what's wrong, or at least state that you can no longer reproduce the error, it does no longer apply, etc.?

[peepo (j-chetwynd)]

#3 Helder,

from what I remember in May...

recent builds 'claim' support for SMIL, even though this is partial.
I proposed removing SMIL support claim whilst support is partial, this was rejected.

the responsible developers believe there efforts are best directed to completing support asap
in the meantime fakesmile is disabled, due to 'claim'

end result is sites remain broken in the interim.
this could be days, weeks, months or years.

regards

tested links today with latest nightlies on OS X ppc and can confirm this is the current status to the best of my knowledge.

[Fenring (leunen-d)]

Hi Jonathan,

You can try to remove the test in fakesmile, and run it whether the UA claim to support SMIL or not.

Or find a test that works for you. (a more elaborate one that tests functionality instead of features string, or one base on UA-string, but i don't advice you that)

David

[codedread (codedread)]

I kind of forget - what is the test in FakeSmile for SMIL support?

I believe Firefox 3.6 and 3.7 should have disabled support for SMIL now. The nightlies will have SMIL support enable-able using the preference, I believe.

[Fenring (leunen-d)]

The tests are on lines 98-103.
document.implementation.hasFeature(...

[codedread (codedread)]

yeah that sucks if what peepo says is true - Firefox reports support for the feature even though they have only minimal partial support?

Either way, I don't think this is FakeSmile's bug.

[peepo (j-chetwynd)]

I regularly raise this type of issue, not only with developers, but the specification Working Groups and even process WG.

It's not a fakesmile-only bug, but there is little point in authors using Fakesmile,
or at the very least they need to be aware of the issues.
which is difficult...

hence I raised this bug to publicise, and raise comments.
it can be closed, but the likelyhood is that someone else may either re-file,
or search harder and maybe in vain for an explanation...

it's a process bug, Fakesmile is part of the process, Fakesmile developers failed to understand, and ensure that there work had a life whilst partial support was.

regards

[peepo (j-chetwynd)]

alternative is to file a bug to enable, whilst partial support.

frankly I dont think the spec will support this, ie the work is not justified.

and it's irresponsible for authors to just assume no support without checking
so imo there probably is no reasonable workaround.

but a bug nonetheless.

[Fenring (leunen-d)]

Gecko's developers closed your bug report stating it's normal for nightlies to contain partial support, experimental features or bugs.
And they're right : people using nightlies are warned that all sites might not work as expected.
FakeSmile intents to work around implementations that support svg but not smil, not buggy smil implementations.
Thus I'm closing this report.

Yet, if you still want to work around that precise buggy implementation, as stated above, I suggest you either :
- sniff for the buggy nightly and bypass the hasFeature test. It's dirty but simple.
- or you can write a more elaborate testcase that actually tests if the smil feature you want to use is correctly implemented natively. More complicated, but it's actually preferred.

cheers.

[peepo (j-chetwynd)]

#11

fenring, fully aware of the issues you raise, my concerns are these:

is this an unnecessary burden for authors?
of what benefit to ua developers?

might this enter the stable release?
  even if unintentionally?

[longsonr (longsonr)]

What peepo says in #4 is not true. Recent Minefield builds do NOT claim to support SMIL as it is incomplete. Nevertheless they DO support some SMIL and this is the problem. What peepo wanted was for us to claim SMIL support with our incomplete implementation, this request was rejected.

[peepo (j-chetwynd)]

longsonr,

I believe you've misunderstood the process and misrepresented my suggestion at the time, iirc, if the application claims to support any SMIL, then FakeSmile is disabled, there is no possibility to support partial SMIL, and for this reason I suggested that as FakeSmile is reliable, until SMIL was more fully implemented, it could be disabled.

However as I have not tested Mozilla for some time, I cannot comment on the current status.

[peepo (j-chetwynd)]

longsonr,
please read through all the above comments, particularly #7 & #8 which better explain the situation as it was a few months ago.

[longsonr (longsonr)]

I did read all comments. As I said before: The application (i.e. mozilla) does NOT claim to support any SMIL. FakeSmil therefrore runs and so does native SMIL and you get a problem. If you want to disable native SMIL then one of the comments in the mozilla bug tells you how to do so.

#8 is incorrect.

[peepo (j-chetwynd)]

fine, my apologies, fakesmil is broken because mozilla wont claim support...

hasFeature isn't going to be that useful to developers in the future as a concept, with this type of deliberate obstruction.

it would be kind if you could take the time to explain why it is important to mozilla not to claim support, nor disable partial SMIL support, when just for now, fakesmil is such a great implementation.

I'd hope that applications could be less monolithic, and this does require cooperation.
why do we need a spellchecker for each and every application that can edit?

[codedread (codedread)]

Jonathan, Fenring said it 4 months ago: "it's normal for nightlies to contain partial support". Nothing has changed with respect to that.

Robert/Daniel, my expectation is that Firefox 3.7 (or whatever it's going to be called) will turn on the feature flag for SMIL support before release. Is this a reasonable expectation? WebKit did this (as can be seen here: http://codedread.com/svgtest.svg) and they don't support 100% of SMIL but enough to make it somethign that developers can use.