W3C home > Mailing lists > Public > public-respimg@w3.org > October 2013

Re: What do we do with picture?

From: Anselm Hannemann <info@anselm-hannemann.com>
Date: Thu, 17 Oct 2013 20:14:12 +0200
Cc: public-respimg@w3.org
Message-Id: <FBAA3F0A-535A-44AE-8A81-D07240F792FF@anselm-hannemann.com>
To: Marcos Caceres <w3c@marcosc.com>
Marcos, answering your questions on my throughts below. Hope it helps to understand my point of view :)

On 17.10.2013, at 17:04, Marcos Caceres <w3c@marcosc.com> wrote:

> On Thursday, October 17, 2013 at 3:41 PM, Anselm Hannemann wrote:
> 
>> On 17.10.2013, at 16:28, Marcos Caceres <w3c@marcosc.com (mailto:w3c@marcosc.com)> wrote:
>> 
>> I personally think <picture> still is the way more intuitive markup for normal developers.
> Could be. But they are equivalent syntaxes used for both. The only different is the use of an attribute.

No, it is the difference to put multiple attributes into one. In picture you could always use src and media attribute and use the verbose syntax.
More advanced users could always use the more advanced compressed syntax of srcset. So while this is easy to teach (agree here with Aaron),
this also serves the other approach with srcset.

>> Some people might remember the approach I proposed 2.5 years ago to the whatwg - it was quite similar to srcN but was rejected by RICG and not followed along by WHATWG either as it has some caveats:
>> 
>> - not easy to understand
>> - harder to maintain due to the {n} number, like removing number 2 out 3
> 
> do you have a pointer?   

Sure, although it's hard to follow the conversation as the subject was changed and there are months between it.
Also I don't link the cg's wordpress discussion here.
This is the first thread about it: http://lists.whatwg.org/htdig.cgi/whatwg-whatwg.org/2011-August/thread.html#32977

> 
>> This said, your points on the complexity of implementation are all true.
>> I hope all of you are aware that this srcN or whatever solution we come up are a milestone and will hold for at least several years.
> 
> Yes, this is an important point. But src-n didn't come out of the blue. Both John and Tab have been very much involved in trying to solve this problem for a long time. These guys have given this stuff as much thought as we have and they have a darn good spec to boot.

Marcos, be sure I honor each effort to come up with a webstandard solution for a valid use-case. But that alone does not mean it's ideal.
Many people have given much thought on various respimg solutions. At least, it's not our point to discuss such things here, it does not matter who put how much effort in what solution  the best solution should be prioritized. For me srcN simply isn't for the named reasons.

>> I simply want to avoid that all developers will complain about this in one year and have to deal with a syntax that people simply don't understand.
> 
> You are talking about the viewport-url? If so, that's just one of the three possible options. I'm still skeptical about viewport-ulr, but I like that media queries and the srcset syntax are both supported (so it's not different from picture, which should hopefully overcome a lot of learning difficulties).  

I am thinking about the srcN attribute itself as well as the viewport things. The maintainability problem of srcN is not deniable IMO: src1="" src3="" and the next dev seeing that wonders where src2 is. It's just feels not right to me (My original proposal for respimg btw did it the same way and I figured this to be wrong). 

>> And I also think we already created a big mess due to the implementation and adoption by developers of srcset which now should be replaced by srcN.
> 
> There are no browsers shipping with srcset so I'm not sure what you mean? Seems that most sites that are using srcset are using it with picturefill.  

Well, WebKit implemented it. And also with picturefill this is a syntax people now adopted.
However, if srcN would be so much metter than srcset and picture I wouldn't mention this but as it isn't in my opinion I hope other users and browser vendors don't get pissed by such changes and don't care about web standard(!) solutions at all anymore. This is what I fear of this big change.  

> Those sites in the top 50k are, according to webdevdata's sept data set:
> 
> vogue.es
> wmich.edu
> marefa.org  
> bitcoin.it
> sportswiki.ru
> cato.org
> 
> Of course, there are more.  
>> Sorry, this might sound a bit harsh but many people don't care about a standard solution of respimg anymore because of all that confusion. And I could
>> imagine that WebKit might be more cautious to implement such new things in future.
> 
> No, I think those assertions are fair. Community-driven standardization is a new thing in our industry and we were bound to make mistakes along the way (we fail better every day - and getting all the browser vendors together in Paris was a testament to that!). I can also understand that people are experiencing standardization fatigue, but we are not doing too bad. The IEEE did a study about the average length of standardization and they found it takes about 5 years [citation needed] so we are pretty much still on track.  

Well, I do think all doing their best effort. But I do understand people why they don't want to take part of webstandard discussions.
And sometimes I can't help but have the feeling that if some specific people give a statement, this is god's given words and has to be right. This is what I really don't like about the whole standard thing but anyway, that should also not be part of our discussion here. ;)
Received on Thursday, 17 October 2013 18:14:41 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 20:12:40 UTC