[widgets] default start file table vs. src attribute

2009-11-20 Thread Cyril Concolato

Hi all,

While implementing the required features to pass the tests of the test suite, I was 
wondering if you really want to keep the default start file table. The benefit of 
this table seems to be just avoiding the use of a content element with an src 
attribute in the config file while the drawback is that you have to scan your zip 
file for all files in order. The spec and conformance would be simpler without that, 
without losing features I think. WDYT?

Cyril
--
Cyril Concolato
Maître de Conférences/Associate Professor
Groupe Mutimedia/Multimedia Group
Département Traitement du Signal et Images
/Dept. Signal and Image Processing
Ecole Nationale Supérieure des Télécommunications
46 rue Barrault
75 013 Paris, France
http://tsi.enst.fr/~concolat 



Re: [widgets] default start file table vs. src attribute

2009-11-20 Thread Marcos Caceres
On Fri, Nov 20, 2009 at 4:58 PM, Cyril Concolato
cyril.concol...@enst.fr wrote:
 Hi all,

 While implementing the required features to pass the tests of the test
 suite, I was wondering if you really want to keep the default start file
 table. The benefit of this table seems to be just avoiding the use of a
 content element with an src attribute in the config file while the
 drawback is that you have to scan your zip file for all files in order. The
 spec and conformance would be simpler without that, without losing features
 I think. WDYT?

The content element is not mandatory, so you kinda need the table.
Also, it needs to be specified what precedence a user agent gives to
loading files, which is the second purpose of the table. The last
purpose of the table is to tell a user agent what MIME type to use for
each file type, which is also fairly important.

-- 
Marcos Caceres
http://datadriven.com.au



Re: [widgets] default start file table vs. src attribute

2009-11-20 Thread Cyril Concolato

Robin Berjon a écrit :

Hi Cyril,

On Nov 20, 2009, at 17:58 , Cyril Concolato wrote:

While implementing the required features to pass the tests of the test suite, I was 
wondering if you really want to keep the default start file table. The benefit of 
this table seems to be just avoiding the use of a content element with an src 
attribute in the config file while the drawback is that you have to scan your zip 
file for all files in order. The spec and conformance would be simpler without that, 
without losing features I think. WDYT?


I actually like it, it's one less thing that we need to specify (I was 
unfavourable to making the configuration requires in the first place). I've 
implemented it and it works nicely. Yes, it's a bit of a performance hit but 
it's not so bad and you can cache it easily.

I agree that it's not a big burden. I think it's more a question of taste. I would prefer 
putting more in the configuration than less. What do you mean by you can cache it 
easily ?

Cyril

--
Cyril Concolato
Maître de Conférences/Associate Professor
Groupe Mutimedia/Multimedia Group
Département Traitement du Signal et Images
/Dept. Signal and Image Processing
Ecole Nationale Supérieure des Télécommunications
46 rue Barrault
75 013 Paris, France
http://tsi.enst.fr/~concolat 



Re: [widgets] default start file table vs. src attribute

2009-11-20 Thread Cyril Concolato

Marcos Caceres a écrit :

On Fri, Nov 20, 2009 at 4:58 PM, Cyril Concolato
cyril.concol...@enst.fr wrote:

Hi all,

While implementing the required features to pass the tests of the test
suite, I was wondering if you really want to keep the default start file
table. The benefit of this table seems to be just avoiding the use of a
content element with an src attribute in the config file while the
drawback is that you have to scan your zip file for all files in order. The
spec and conformance would be simpler without that, without losing features
I think. WDYT?


The content element is not mandatory, so you kinda need the table.
Also, it needs to be specified what precedence a user agent gives to
loading files, which is the second purpose of the table. The last
purpose of the table is to tell a user agent what MIME type to use for
each file type, which is also fairly important.

As I replied to Robin, I'd rather make it more explicit in the config file. I 
think it would make it more readable/understandable without requiring to 
know/hard-code all the default values. It's a question of taste. My suggestion 
was more to mandate the content element, to mandate a type attribute and to 
allow nesting them for indicating the precedence or use document order 
(multiple content elements, like the icon element).

Cyril
--
Cyril Concolato
Maître de Conférences/Associate Professor
Groupe Mutimedia/Multimedia Group
Département Traitement du Signal et Images
/Dept. Signal and Image Processing
Ecole Nationale Supérieure des Télécommunications
46 rue Barrault
75 013 Paris, France
http://tsi.enst.fr/~concolat 



Re: [widgets] default start file table vs. src attribute

2009-11-20 Thread Robin Berjon
On Nov 20, 2009, at 18:36 , Cyril Concolato wrote:
 Robin Berjon a écrit :
 I actually like it, it's one less thing that we need to specify (I was 
 unfavourable to making the configuration requires in the first place). I've 
 implemented it and it works nicely. Yes, it's a bit of a performance hit but 
 it's not so bad and you can cache it easily.
 I agree that it's not a big burden. I think it's more a question of taste. I 
 would prefer putting more in the configuration than less.

Then our tastes differ :) I really think that Convention Over Configuration is 
the way to go here (cf 
http://en.wikipedia.org/wiki/Convention_over_configuration). In general, the 
more defaults there are, the less cases of outright failure there are, the 
better. Otherwise it just becomes Java programming all over again, and no one 
wants that, right?

 What do you mean by you can cache it easily ?

You read the widget's configuration once, then when you read the same widget 
again you've cached the configuration and don't need to find the start file 
again (unless the locale has changed).

-- 
Robin Berjon - http://berjon.com/