[UPS] RE: UPS digest, Vol 1 #46 - 1 msg
Carl Lagoze
lagoze@cs.cornell.edu
Tue, 4 Jan 2000 15:25:01 -0500
Simeon,
THanks for you comments. Responses are inserted at appropriate places in the
text.
Regards,
Carl
> -----Original Message-----
> From: ups-admin@vole.lanl.gov [mailto:ups-admin@vole.lanl.gov]
> Sent: Tuesday, January 04, 2000 2:00 PM
> To: ups@vole.lanl.gov
> Subject: UPS digest, Vol 1 #46 - 1 msg
>
>
>
> Send UPS mailing list submissions to
> ups@vole.lanl.gov
>
> To subscribe or unsubscribe via the web, visit
> http://vole.lanl.gov/mailman/listinfo/ups
> or, via email, send a message with subject or body 'help' to
> ups-request@vole.lanl.gov
> You can reach the person managing the list at
> ups-admin@vole.lanl.gov
>
> When replying, please edit your Subject line so it is more
> specific than
> "Re: Contents of UPS digest..."
>
>
> Send UPS mail list submissions to ups@vole.lanl.gov
> To subscribe or unsubscribe via the web, visit
> http://vole.lanl.gov/mailman/listinfo/ups
> You can reach the person managing the UPS list at
> herbert.vandesompel@rug.ac.be
> _______________________________________________
>
>
> Today's Topics:
>
> 1. Re: Open Archive Software (Simeon Warner)
>
> --__--__--
>
> Message: 1
> Date: Mon, 3 Jan 2000 13:26:13 -0700 (MST)
> From: Simeon Warner <www-admin@arxiv.org>
> To: <ups@vole.lanl.gov>
> Subject: Re: [UPS] Open Archive Software
>
>
> To Carl and cc: to everyone else,
>
> The following are comments on the Open Archives Dienst subset
> defined in:
> http://www.cs.cornell.edu/cdlrg/dienst/protocols/OpenArchivesD
> ienst.htm
> (Document Version 0.3 1999-12-22 15:44:37 -0500 )
>
> General:
> OAMS or oams? -- all the OpenArchives pages refer to OAMS
> yet the Dienst
> docs refer to oams.
oams or OAMS really doesn't matter to me. If it is important to people, I
can make the protocol have it as OAMS but it really doesn't make much
difference. In general, in Dienst we have kept these things to lower case.
>
> Disseminate:
> Why #format? not just format?
Because the Open Archives Subset is indeed a subset of the larger Dienst
protocol with considerably more functionality attached to the Disseminate
verb. See
http://www.cs.cornell.edu/cdlrg/dienst/protocols/DienstProtocol.htm#Reposito
ry-Disseminate and you'll notice that the 'meta-format' fixed argument in
the subset actually corresponds to the 'view' fixed argument in the full
protocol, which takes a number of forms to allow flexible dissemination from
a document. Sorry, but we need to maintain compatibility between these
protocols since full Dienst sites should also be OA compliant.
>
> The example (according to spec.), should be
> Dienst/Repository/1.0/Disseminate/handlecorp%2f970101/%23oams/xml
> instead of
> Dienst/Repository/1.0/Disseminate/handlecorp/970101/%23oams/xml
>
> HOWEVER, it actually makes no difference because the CGI
> specification
> says that the PATH_INFO part of a URL should be decoded (ie %2f to /)
> before being passed to the script. See:
> http://hoohoo.ncsa.uiuc.edu/cgi/env.html
> Also, noted by Apache at:
> http://bugs.apache.org/index/full/876
>
> I point this out because it means that fixed args in Dienst do
> not nicely allow `/' and other escaped characters through.
> Since arXiv
> identifiers contain a `/', I got caught out by this one. Keyword
> args (anything after `?') are passed un-decoded in QUERY_STRING.
Hmmm, it should be '/' and not the encoded version especially since Apache
gets upset at encoded '/', doesn't it? For example, try:
http://www.ncstrl.org:8070/Dienst/Repository/1.0/Disseminate/xxx/001/%23oams
/xml.
Your comment that fixed args in Dienst don't nicely allow '/' but we would
have run into that problem with just about any character. If you take a
look at the code in our reference software at
http://www.cs.cornell.edu/cdlrg/dienst/software/openarchives.tar.gz you'll
see how we handle this (in the file Dispatch.pm). Please let me know if
this clears this up.
>
> List-Contents:
> Inconsistent use of partition/partitionspec
Whoops, thanks for catching that one. Damn hard to keep every detail under
control in these types of documents. Actually, the version number was
wrong, should be 4.0 rather than 3.0 (as indicated in all the examples).
>
> Should there be a facility to list a set of partitions rather
> than just one (or by default, all)?
In other words, you'd like an argument that specifies the root of a true
under which you'd like to display the partition structure? Let me know if
this is what you want and I'll check if there any problems with my
programmer. Shouldn't be any, but you never know.
>
> Here the meta-format option is inconsistent with Disseminate because
> there is no `#' before the format name.
By here, I assume you mean within the List-Contents verb. This is because
the value is not overloaded with other meaning in the full protocol (as
described above).
>
> The return format seems strange with the fullId put after the record
> token:
>
> <List-Contents version="3.0">
> <record>
> handlecorp/970101
> <rfc1807:rfc1807
> xmlns:rfc1807="ftp://nic.merit.edu/document/rfc/rfc1807.txt">
> <rfc1807:author>William Shakespeare</rfc1807:author>
> ....
> </rfc1807:rfc1807>
> </record>
> ....
> </List-Contents>
>
> It would make more sense to me to have:
>
> <List-Contents version="3.0">
> <record fullId="handlecorp/970101">
> <rfc1807:rfc1807
> xmlns:rfc1807="ftp://nic.merit.edu/document/rfc/rfc1807.txt">
> <rfc1807:author>William Shakespeare</rfc1807:author>
> ....
> </rfc1807:rfc1807>
> </record>
> ....
> </List-Contents>
>
> which is then similar to the structure of the
> List-Partitions response.
Ah yes, the argument over attributes vs. entities in XML DTDs. We've had
some religious wars over that here and my two XML people seem to insist that
we've got it right. I can't seem to recall the exact justification for this
DTD but I prefer to stick with it now unless you have radical objections.
>
> Cheers,
> Simeon.
>
>
>
Thanks again for the close look and please get back to me if you have
further comments or suggestions.
Best,
Carl
>
>
>
> End of UPS Digest
>