Re: Proposed note about C064 and C065

At 23:27 07/02/20, Fran〓ois Yergeau wrote:
>
>Martin Duerst a 〓crit :
>> I'm not very happy about this. I think the intent of the WG at the time
>> of authoring the document was clearly different from the text below.
>
>My recollection is different, but of course that was a while ago and I might be wrong.
>
>> In my understanding, the WG felt that it was important that all specs
>> would use the best available (i.e. newest) references to the Unicode
>> Standard, even if this meant the use of a minor version and longer
>> text in the reference section.
>> I think there were various reasons for this (I don't think they were
>> discussed explicitly):
>> - Contrary to many other specifications, the main point distinguishing
>>   major from minor versions in the case of the Unicode Standard is
>>   whether a book is produced or not. Production of a book is not only
>>   related to content, but also to timelines,... It is therefore easily
>>   possible, and has happened in the past, that a 'minor' update added
>>   important scripts and characters.
>
>This ("'minor' update") confuses minor versions and update versions of Unicode.  Update versions (the 3rd number in x.y.z) do not add characters or scripts and are uninteresting to most specs, especially in the context of a generic reference where one is trying to say "do whatever the current Unicode standard says".  The proposed note does talk about "the very latest update version of Unicode".

So do you mean that other specs should track minor versions (second
digit in version number) but not update versions (third digit)?
If we want to make that distinction, then we have to add a lot more
detail.

>> - It seems that there are some people that have difficulties of getting
>>   the idea of a generic reference.
>
>Agreed. And perhaps our choice of calling it "generic" was not the best one.

I don't think the term is the problem. It's that some people have
problems accepting references to something that's evolving.

>> As an example, the reference to
>>   Unicode 2.0 in XML 1.0 was (except for name syntax) clearly intended
>>   as a generic reference, but many people thought that the spec as
>>   written made it impossible to use post-Unicode 2.0 characters *in data*.
>>   To help such people, even if just a little bit, using the newest
>>   reference may be better. (and of course the "as updated from time to time"
>>   text is even more important).
>
>Of course the "as updated from time to time" is what matters most.  As for using the newest reference, the proposed note does not discourage that, in fact it gives pointers where to find the required info.  What the note makes clear is that there is no obligation to *track* the latest version, as in updating a spec whenever Unicode issues a new update.

I agree that there in no need to track Unicode versions/updates in spec
errata. Richard has proposed text that clarifies this. But this was
not the original issue that Bjoern came up with.

It would, however, in my view be very good if e.g. new Editions of a spec
would update their references. A typical example is XML, which still
references XML 2.0.

>> - It seems unhelpful to the reader of the spec (and remember, there are
>>   usually many more readers than writers) to have to find out on her own
>>   about already available updates. It is clear that we can't list future
>>   updates, but dropping current updates just as a convenience to the
>>   spec writer seems inappropriate.
>
>I don't think that argument holds water.  In all cases the reader does have to check, there's no other way to find out whether there have been updates since the publication of the spec, whatever the starting point is (current at time of publication or not).  This is just a fact of life with a version-independent reference to a standard that does have new versions from time to time.

Well, yes, but it translates into: "The reader has to check anyway,
so why would I care." for the spec writer, which then easily can
be read by the reader as "Whoever wrote this spec didn't care, so
why should I?". That's definitely something I'd want to avoid.

>> The above reasons, and the fact that the WG itself carefully made sure
>> that the full version (including minors) was in the reference seem to
>> support this. If we want to change this, I suggest that:
>> - We preferably do so in a general update of Charmod.
>> - We actually change the examples to match what the spec says.
>
>In what way?  The spec says to use the latest version and the examples do that (but of course are stale by now).  What should we change here?

Yes, but the Note says that it's not really necessary. Saying
"Reference the latest version, but you don't really have to do so."
is a sign of a bad spec. If we change to "Reference the latest major
(and if availible minor) version", that's fine, but in that case
we better change the actual spec text, and the example, rather than
muddying things with a note.

(Note that C064 says: "All generic references to the Unicode Standard
[Unicode] MUST refer to the latest version of the Unicode Standard
available at the date of publication of the containing specification.",
and C065 likewise. I don't see any way to interpret "latest version"
other than including minor and update versions.)

Regards,     Martin.



#-#-#  Martin J. Du"rst, Assoc. Professor, Aoyama Gakuin University
#-#-#  http://www.sw.it.aoyama.ac.jp       mailto:duerst@it.aoyama.ac.jp

Received on Wednesday, 21 February 2007 05:34:25 UTC