[whatwg] Browser Bundled Javascript Repository

[Ian Fette (イアンフェッティ)]

2009/6/15 Joseph Pecoraro <joepeck02 at gmail.com>

> On Mon, Jun 15, 2009 at 1:09 PM, Joseph Pecoraro <joepeck02 at gmail.com>wrote:
>
>> Dion: The problem here is that isn't backwards
>> compatible and thus no-one will really be able to use it.
>>
>>
>> I thought the original idea was backwards compatible. Maybe not the URN
>> Schemes. If the original idea is not, could you point out the issues?
>>
>
> The URN schemes isn't compatible. The SHA hash idea is do-able, but as
> Oliver pointed out is impractical: a) devs will forget to update it, b)
> looks ugly, c) fun things would happen with a SHA collision! ;)
>
>
> a) Solved by Validation - I can't think of anything much better then that.
> =(
> b) Canonical Listing - This shouldn't be too difficult to distribute from a
> central source or some convention.
> c) Hehe, I think I detect a hint of sarcasm.  If there is a SHA1 collision
> then you'd probably make a lot of money!
>
>
C is a serious concern. SHA-1 collisions are now 2^51 -
http://eprint.iacr.org/2009/259.pdf



>
>
>  Dion: You then also get into the "how do I get my library into the browser?"
>
>>
>> Enough widespread usage of a library is a clear indicator for adoption
>> into a browser bundle.  Dynamically growing repositories could optimize per
>> computer for the particular user's browsing habits (assuming developers
>> would mark their scripts with the identifiers).
>>
>> You can have the same problem with what libraries will Google include in
>> its CDN.  Although it may be easier for Google to host just about any
>> library if it already has a CDN setup.
>>
>
> This was a real problem for us. How much is "enough" ? We started to get
> inundated with requests for people to put libraries up there.
>
>
> Lets the browsers decide.  And I can't make any reasonable suggestions
> without getting real world data, something I haven't tried to do yet.  But
> yes, this is a good point, something that is extremely flexible / variable.
>
>
>  Dion: After mulling this over with the Google CDN work, I think that using
> HTTP and the browser mechanisms that we have now gives us a lot without any
> of these issues.
>
>>
>> I was afraid of this.  This is a completely valid point.  I guess it
>> sounds like too much work for too little gain?
>>
>
> I don't want to stop you from working on these ideas. The core problem that
> we tend to download the same crap all the time is real, and I look forward
> to seeing people come up with interesting solutions.
>
>
> Thanks for the support.  My thoughts are beginning to look like this:
> - Javascript Frameworks are downloaded all the time on many domains. This
> is a special case.
> - Those who benefit the most are the ones that can't space the extra
> request or large caches.  This makes me think mobile browsers would get the
> biggest benefit.
> - I think the iPhone had some special html syntax for its mobile webpages,
> maybe they can sneak this in if it proves useful to them.
>
> - Joe
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.whatwg.org/pipermail/whatwg-whatwg.org/attachments/20090615/23e56415/attachment-0002.htm>