ietf-nntp Fwd: HDR/OVER LIST OVERVIEW.FMT thoughts

[Russ Allbery]

Jim Calvin <jim at jced.org> wrote:

> What's possible should be neutral to any implementation - possible but
> it's usage might make some server admin unhappy. As an example, there
> may be implementations for which getting to the OVERVIEW database is no
> better than parsing the article (unlikely perhaps, but possible).

I'm not too worried about the overview database, but with HDR, there
definitely can be a large difference in the efficiency of retrieving some
headers compared to others.

> Perhaps LIST EXTENSIONS should have a means to say explicitly which
> header fields are "efficient" for the server. In part that's what LIST
> OVERVIEW.FMT probably did at one time. But at this point, it seems a
> little too tied to a particular implementation.

This is something that could be done with the LIST HEADERS proposal, as a
second field for those headers that are considered efficient.  Do people
feel that it would be worth the additional complexity?

It's not clear to me how best to make that interact with HDR ALL.  One
approach would be the one below:

> LIST HDR.AVAIL provides what is accessible via HDR in the form of

> header1~
> header2~
> header3~
> ALL

> Where "~" (or some other token) indicates "this is efficient for this
> server to retrieve." The ALL token says the server will retrieve any
> header that can be found. The example above says that the server will
> retrieve any header, but header1-header3 are efficient.

-- 
Russ Allbery (rra at stanford.edu)             <http://www.eyrie.org/~eagle/>