Dear Adam,
Many thanks. All the links to the source code you gave me helps me a lot to
understand the situation much better. A pity this is so elaborate.
In the meantime our case with Thomson Reuters resulted in directing us to a web
site, of which the answer e-mail claimed, it would supersede the sites we
discussed previously. Here is that URL, since it may also be of interest:
http://wokinfo.com/products_tools/products/related/webservices/
Despite all being very helpful what you wrote, it is certainly not good news.
And I share several of your concerns only too easily. I am willing to help. My
first step along this lines is currently to continue insisting that the company
really helps us. My institution is paying AFAIK huge amounts of subscription
fees, yet nobody seems to have informed us of the consequences of the
transition as end of August nor do they really help at the moment to overcome
the difficulties. The information they have provided so far are all very
conflicting and no real progress can be seen. Obviously as an end-user one has
to insist they help, given the way they have handled us during this recent
transition, or we will never get anywhere. The case only exists, because I
complained insistingly, which finally did trigger this case and so far a few
reactions.
However, would it not be easier not to work with Mountain Lion first, in
particular to overcome the caveats of the new protocol? Say, using the reliable
Snow Leopard (or Lion, but I wouldn't use the word reliable in this case ;-) ).
Once understood it might work even under Mountain Lion?
One further thought: Such SOAP like protocols
(http://en.wikipedia.org/wiki/SOAP_(protocol)) may be of interest also in other
cases. Would it not be more robust for BibDesk to develop a generic module that
can handle such queries while the syntax specifics of any given protocol are
described with a meta language, e.g. similar as attributed EBNF
(http://en.wikipedia.org/wiki/EBNF) this can do for programing languages. Then
BibDesk could be extended by any query by simply providing the protocol
description in form of that language? What is your experience with these
interfaces? Are they clean? Formally well designed or are they all just
tedious, cumbersome patchwork full of idiosyncrasies?
Regards,
Andreas
On 11/09/2012, at 01:17 , Adam R. Maxwell wrote:
On Sep 10, 2012, at 14:48 , Fischlin Andreas
mailto:andreas.fisch...@env.ethz.ch>> wrote:
Dear Adam,
Many thanks for this most valuable information. Indeed, in the meantime I got a
confirmation from company Thomson Reuters that they have switched from version
1.0 to 2.0 of their web services, including the SOAP access as of 31st August
2012.
Well, that's a pretty clear indication of what has happened.
Unfortunately I need very urgently ISI WOS in BibDesk back.
Any idea what could be done?
New WS stub code [1] has to be generated with WSMakeStubs, it has to be checked
for bugs, and then glue code has to be written to integrate the new service
into BibDesk [2]. This last is the only significant task, and based on my
cursory glance last week, it'll be a fairly extensive rewrite.
BDSKISIWebServices.[hm] will be replace entirely, and BDSKISIGroupServer.m will
have to be rewritten to use the new methods and error checking.
[1]
http://bibdesk.svn.sourceforge.net/viewvc/bibdesk/trunk/bibdesk/BDSKISIWebServices.m?revision=18776&view=markup
[2]
http://bibdesk.svn.sourceforge.net/viewvc/bibdesk/trunk/bibdesk/BDSKISIGroupServer.m?revision=18776&view=markup
I'd guess 4-8 hours work, assuming the new API is straightforward and not too
different from the old one, but I have very low confidence in those
assumptions, and I have no time to look at this myself before October.
Notably I have for sure following questions for the time being:
1) Where are search details as contained in a .bdsksearch file actually stored
in BibDesk once such a search has been installed?
I have consulted the help pages, and examples of .bdsksearch files are also not
that difficult to construct from scratch. But this all did not answer me this
question. Have I overlooked something or is it really missing.
Those files are largely irrelevant for WoS searches; they just store a database
name and human-readable counterpart. Local additions should end up in
~/Library/Application Support/BibDesk/SearchGroupServers.
2) What URL is "Searches -> Web of Science SCI" using?
In the current case I opened with Thomson Reuters, they are asking me this
question, but I do not know what to answer.
The default endpoint is
http://wok-ws.isiknowledge.com/esti/soap/SearchRetrieve
See link [1] for details.
3) Can I edit or at least see Web of Science SCI search specs at the XML level?
Is there a way to edit a search once its .bdsksearch file has been installed? I
looked for plist files, but neither inside the application BibDesk nor in
~/Library/Application Support could I find the wante