Re: Warning text at odds with Mobile Web Best Practices

Hi,

Many thanks for the report, Paschal!

I agree with you. The test actually does not detect incorrect uses of 
caching techniques, but rather:
  1. the absence of caching information (which would trigger a failure)
  2. the presence of caching directives that prevent caching (warnings)
  3. a few cases where the server does not return an optimal response 
when "Last-Modified" or "ETag" are used (warnings as well)

The purpose of most of these warnings is not say: "you've done something 
wrong", but "make sure that prevented caching on purpose because it has 
a particular impact on mobile networks".

In particular, there is no attempt to warn about the fact that the 
caching value may not be appropriate for the content in itself.

I softened the wording of the test explanation to:
"Detects lack of caching information and possible improvements [...]"
... and completed the warning information with a link to a tutorial on 
caching.

What is still missing IMO, and I'll add it in a later phase, is a global 
explanation that warnings are usually "neutral", i.e. may or may not 
have to be addressed, depending on the context.

Francois.


Abel Rionda wrote:
> Hi Paschal,
> 
>> The warning says "Detects incorrect uses of caching techniques"  -
> 
> Yes, perhaps it is a bit severe to say it in this way since it is just a
> warning. Something like the text you propose would be ok. However, I
> think this output is part of the checker interface and not of the
> library itself,
> so perhaps Francois or Dom can help with this.
> 
> 
>> Section 5.4.15.2 says that it is perfectly correct to use nocache to
>> prevent caching.
> 
> The text you mention simply states that you should consider using these
> headers depending on your application needs.
> Anyway, the checker code implements mobileOK Basic Tests 1.0, which are
> machine testable algorithms based on some of the Mobile Web Best
> Practices 1.0. If you review Caching test [1] you can see that these
> warnings are aligned with the document. 
> 
> Regards,
> Abel.
> 
> [1] http://www.w3.org/TR/mobileOK-basic10-tests/#CACHING
> 
> -----Mensaje original-----
> De: public-mobileok-checker-request@w3.org
> [mailto:public-mobileok-checker-request@w3.org] En nombre de Paschal Nee
> Enviado el: martes, 09 de diciembre de 2008 14:53
> Para: public-mobileok-checker@w3.org
> Asunto: Warning text at odds with Mobile Web Best Practices
> 
> 
> Hi,
> 
> The warning provided when a page has no-cache headers set is at odds
> with the Mobile Web Best Practices 1.0 section linked from the warning
> itself.
> 
> The warning says "Detects incorrect uses of caching techniques"  -
> incorrect being the key word here
> Section 5.4.15.2 says that it is perfectly correct to use nocache to
> prevent caching.
> 
> An example of what I am talking about can be found here:
> http://tinyurl.com/w3c-cache
> 
> Personally I do not like this warning in the first place.  A cache
> control header of public or private can be equally incorrect depending
> on the page in question.  But there is no warning if the cache control
> is set to public or private.
> 
> At a minimum I think the language in the warning should be change to
> reflect the reality e.g. something like removing the word incorrect
> and adding "This practice may be completely applicable depending on
> the needs of your application"
> 
> Thanks,
> Paschal.
> 
> PS - I'm not subscribed to the list so please reply to all if you need
> further input from me.
> 
> 
> 
>

Received on Wednesday, 10 December 2008 22:26:02 UTC