http://gcc.gnu.org/onlinedocs/libstdc++/ext/lwg-defects.html
Jonathan Wakely
jwakely.gcc@gmail.com
Mon Apr 11 19:11:00 GMT 2011
On 11 April 2011 17:38, Paolo Carlini wrote:
> Hi,
>>
>> If these are just copied from the original site, can we please remove
>> that local copy? More often than not we've not been up-to-date,
>
> Well, normally we are up to date. For us being up to date means that all the
> entries in intro.xml point to the corresponding entry in one of the
> lwg-*.hml files in the form corresponding to our actual implementation in
> v3. If something happens at the ISO level about one of those entries *and*
> that is reflected in our implementation normally we import new lwg-*.html
> files.
>
> Anyway, I didn't invent the concept of importing lwg-*.html files. I think
> the idea dates back to when people where supposed to download a v3 tarball
> and find it as self-contained as possible, without the need to access the
> internet while looking at the defects. If the other maintainers caring a lot
> about docs, eg, Benjamin, Jonathan, are Ok with the change I don't have a
> strong opinion either way. Note, however, that I can see at least a negative
> consequence: if the ISO web pages change, eg, often Defects are shuffled
> around from one lwg-*.html to another, then we have dead link in our
> intro.xml. Today this is very unlikely to happen, modulo stupid errors at
> import time (personally, I always carefully go through all the links in
> intro.xml and make sure that problem vs our imported files doesn't manifest
> itself, I manually adjust the links in intro.xml for that).
Yes, I think it was useful including the defect list in the v3 tarball
before always-on internet connectivity was common, but it's less
important now.
To avoid broken links when issues move between lists, we could link to
a specific N-numbered document in a mailing, instead of the latest
lwg-*.html pages, although that would require more effort to change
every link when we want to refer to a newer revision of the list.
More information about the Libstdc++
mailing list