Re: comment on TC documents

[Lois Goldthwaite <yyyy@xxxxxxxxxxxxxxxxxxx>]

Well, you shouldn't place too much emphasis on interim documents 
produced at various stages in the project. Certainly an intermediate 
draft should not be quoted as authoritative, because it may have changed.

The "list of changes" I was referring to may not have a document number 
yet, but it consists of 256 (!) pages. However, this is because it does 
not take the usual form of "on page 23, paragraph 4, sentence 2, change 
'must' to 'shall'". Instead it shows the complete context of all 
changes, with complete "before" and "after" versions of each paragraph 
changed.

The reason I mentioned the C++-TC-as-integrated-document is because this 
appears to be slightly controversial. The C++ committee was not sure 
whether ISO would even accept a TC in this format, rather than a minimal 
list of changes. What tipped the balance was when we found out that at 
least one national body (the UK) has a long-standing policy of 
integrating TCs on their own and publishing amended documents.[1] When 
they issue these as electronic versions, they are obliged to insert TIFF 
images of the amended pages into the PDF files, which means that text 
search and cut-and-paste silently do not work on those pages. Also, the 
committee preferred to have the integrating done by the C++-literate 
project editor rather than the staff at BSI.

Lois

[1] The reason no one realised this for years is because BSI charge so 
much for their standards documents that nobody buys them!

Alexander Terekhov wrote:

> Lois Goldthwaite wrote:
>
> > FWIW, the C++ standard's recent TC takes the form of a complete new
> > document with the changes integrated. But then our list of changes was
> > NOT a short html document. :-)
> Yeah, "FWIW"(*).
>
> http://std.dkuug.dk/JTC1/SC22/WG21/docs/papers/2001/n1316
>
> This one still works (until tomorrow, I guess), but is
> ~85000 bytes smaller (body.pdf) than the one that I've
> downloaded a couple of months ago -- probably the stuff
> that is now "available" here:
>
> http://std.dkuug.dk/JTC1/SC22/WG21/prot/14882fdis/n1334
>
> regards,
> alexander.
>
> (*) http://groups.google.com/groups?selm=3D6E20B4.6F855683%40web.de
>
> "...
>  standards committee have ALREADY legislated and proclaimed
>  that the standard C{9899:1990,9899/Amd.1:1995} library (a
>  subset of the standard C++ library) SHALL NOT THROW[2]:
>  ...
>  [2] That's really "kinda funny"... given that quite a few
>  *standard* C library functions HAVE BEEN DECLARED TO BE
>  THREAD CANCELLATION POINTS BY >>THE POSIX STANDARDS
>  COMMITTEE<<..."