W3C home > Mailing lists > Public > public-svg-wg@w3.org > October to December 2008

Re: Problem with coords-viewattr-05-t.svg in Tiny 1.2 test suite

From: Alex Danilo <alex@abbra.com>
Date: Sun, 12 Oct 2008 19:36:57 +1100
Message-Id: <LLAM8K.XMDKL3LOB7Y5@abbra.com>
To: cyril.concolato@telecom-paristech.fr
Cc: public-svg-wg@w3.org

Hi Cyril,

--Original Message--:
>Hi all,
>It's been a while since I did not send email on this list. I'm trying to become a little bit more active. Let me start with my 2 cents on this issue.
>The spec says that the type attribute is a hint. As a hint it can be ignored, right? Therefore sniffing is allowed, no? It may be a bad practice according to the TAG finding but it is not forbidden.
>Additionally, the spec says "the server metadata is authoritative over the type attribute". So when loading a file from the system, without real client/server communication, one can consider that the OS is the server and that the file extension is some sort of metadata and therefore displaying the image.

I fully concur with the intention of your comments, but please
see below.

Firstly, the 'type' in question is not an attribute, and therefore
not necessarily a hint. It's part of the 'data:' URI field which
is covered by a specific RFC. SVG doesn't control what the
existing RFC says.

The fact is that the content is inline, not fetched
from a server, namely being from a 'data:' URI. So there
is no type negotiation done, it's an inline part of the
SVG document, thus no server metadata.

Also, the error is quoting 'image/jpg' as the MIME type which
is incorrect and the correct type being 'image/jpeg' as specified
in RFCs which are more than a decade old.

There is no file extension referenced as part of the test error.

Thanks for the comments, but the content of the test should
be correct.

Also, as an aside, as far as sniffing goes - ASV sniffs all
'data:' URIs and does not need the 'type' at all. This is
in violation of the BNF for the 'data:' URI as per its RFC.

My complaint was about the correctness of the test only.


>Alex Danilo a écrit :
>> Hi group,
>> 	At risk of sounding like a broken record, the test-fest shows failure
>> for a couple of implementations and passes for most implementations for the test
>> 'coords-viewattr-05-t.svg'.
>> 	I know I've said it before, but since no-one seems to have paid
>> attention: THE INTERNET MEDIA TYPE (a.k.a. MIME TYPE) IS WRONG. Will someone
>> please listen this time, please...
>> 	All the referenced 'data:' URIs have the media type of 'image/jpg'.
>> The correct media type is 'image/jpeg'.
>>                                    ^
>> 	Media types aren't 3 letter file extensions. As such, all implementations
>> marked as a pass should be marked as fail, for magically deducing the type from
>> an incorrect media type - possibly sniffing content as ASV does, or whatever.
>> 	The corrected test works fine in one failing implementation that
>> I've checked.
>> 	Can we please get this fixed? I'm sure this is the third time
>> I've raised it over the years.
>> Alex
>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
Received on Sunday, 12 October 2008 08:37:36 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:20:09 UTC