Re: Which output version is considered the "canonical" validator output?

Myself wrote:
> 
> Most specifically, I want to implement messageid notations in the SOAP 
> output (I notice Miguel Gastelumendi requested the same thing a few days 
> ago), but the perl templating system is not letting me get very far. 
> While I have a good familiarity with perl, the templating system is 
> resisting being cracked (so far).
> 
> I was going to "settle" for using the SOAP output as-is, but I've run in 
> to an issue tonight.
> 
> In analyzing the following URL, the HTML, SOAP and XML outputs all 
> return different information using the default settings on:
>    http://ebni.com/byrds/
> 
> - The HTML output has a single fatal warning(04):
>    No Character Encoding Found!
> - The XML output has 70 warnings (it must be assuming a char encoding)
> - The SOAP output just appears to be broken
> 
> Is the HTML output considered to be the "canonical" version? If so, I 
> could pare down/demolish the HTML templates to generate exactly what I 
> want (which is pretty close to the SOAP output, but if I can't trust its 
> output compared with the HTML version, I'll take what I can get).

I answer my own email here at the risk of seeming a little repetitious, 
but this day has yielded some progress on my issues. Maybe someone with 
a similar problem will glean something from my stumbling around.

* Regarding the XML output, as stated in xml_output.tmpl:
    "This interface is highly experimental and the output *will* change
    (probably even several times) before finished. Do *not* rely on it!
     See http://validator.w3.org/docs/users.html#api-warning

So, that is *definitely* not canonical.

* SOAP is not canonical either, because fault reporting in the template 
mechanism (soap_fault) appears to be not working at all. The code in 
soap_fault.tmpl seems correct compared to the code in fatal-error.tmpl 
for documenting fatal errors...it just isn't reporting anything. This is 
probably bug 4093 or 4489. I'm hazarding a guess that the 
abort_if_error_flagged routine in the "check" perl script is where the 
magic should be happening. A key statement here at Line 2293 says, 
"@@FIXME: This is borked after templatification." Hrm.

Not canonical either then.

Not to say that I expect the validator output to be totally bug-free, 
but I'd like to have all errors and warnings firing on all cylinders.

So, I'm left with treating the HTML output as being the "canonical" 
output. I'm OK with that.

Lastly, I've made progress in adding error message ids to the output 
stream, both in the HTML version and the SOAP, but doing so was part of 
a destructive process since I didn't want to keep any of the 
human-readable messages. I doubt it would be of interest to anyone else 
but if there is, I'd be happy to share. Unless this whole thing is an 
old issue - in which case never mind. =)

HTH,
-Brian

Received on Wednesday, 16 May 2007 05:54:29 UTC