[cc'ed the list - please send your replies to the list]
Hi,
you've mentioned a couple of problems in your mail. It would be easier
to look at one at a time, but I'll simply comment on what you wrote:
Damien Jade Duff writes:
> I am not sure whether I should rather be sending this to your mailing list.
> I have done some testing on the the latest pre-release of refdb and there
> might be an issue with the RISX - I think it's with the getref part, with
> RISX fromat. The problem existed in the 0.9.5 version though I took this
> upgrade as an oppurtunity to redo my tests without having to learn how to
> retrieve via CVS.
The mailing list is always preferable because others are likely to be
affected by the same problems. Besides, there is a searchable archive
for the benefit of others that may bump into these problems later.
>
> I thank you for any time that you choose to spend looking at this issue.
> Below this message is a printout of my terminal activity, with a printout of
> the refdbd.log at the bottom. I have also available my terminal activity
> during the configure phase, and also the configure.log file which looks
> potentially useful. Please ask me if you need this information or anything
> else, but I hope you'll be able to replicate the issue! If not, I guess I'll
> probably have to try and get my head around C programming...! (sometime in
> the hazy future...)
>
Unfortunately I didn't see the config.log in your mail. Which OS are
you using anyway.
> My main issue is that I would like RefDB to be idempotent with respect to a
> getref/updateref cycle - if I retrieve a reference from RefDB then update it
> back into RefDB without changing it I should hope for there to be no effect.
> This would be useful (nay, essential) if for instance I conduct a query to
> retrieve some XML, only change a part of that XML and then update the whole
> lot back into the DB.
>
This is only partly doable. RefDB does change a few things in your
input data (this is documented in the manual). However, if you
re-import data exported from RefDB, then of course you should get the
same data as before. If this is not the case, as you've seen with your
data, then this is clearly a bug.
> I think this only applies to the RISX format and not the RIS format, though
> - as you can see in the test log below I can still retrieve author and title
> information using the RIS format but not the RISX format.
>
This is true. The data arrive properly in the database, but the risx
backend screws up the data. However, this problem seems to be confined to
CONF entries.
I've attached a fixed version of backend-risx.c. Please use this as a
drop-in replacement for the 0.9.6-pre3 version. I've also attached a
CONF entry that appears to survive import and export properly.
> A note: You may notice some strange behaviour halfway down the test script.
> I get "Parse error at line 1: no element found" when trying to add or update
> XML - but I get it unpredictably - I think you'll probably want to try and
> ignore that as it may be a foible of my system and I doubt it is related to
> the possible bug I described above, which occured in 0.9.5 and I do not
> remember getting the intermittent parse-error behaviour when I was using
> 0.9.5. Also, I don't know if it's a problem with my system or not, but in
> order to prevent the following compile error (once again, only in the latest
> prerelease)...
>
This error message "Parse error ..." is issued by expat, the library
that reads the XML data. Unfortunately your refdbd log apparently does
not contain such a case, so I lack the context in order to see what
went wrong. In any case, it *might* be a problem on your system as I
haven't seen this problem on several installations for years.
> refdbdref.c: In function `read_ris_data':
> refdbdref.c:674: warning: passing arg 2 of `iconv' from incompatible pointer
> type
> refdbdref.c: In function `is_journal':
> refdbdref.c:2395: `ULONG_LONG_MAX' undeclared (first use in this function)
> refdbdref.c:2395: (Each undeclared identifier is reported only once
> refdbdref.c:2395: for each function it appears in.)
> refdbdref.c: In function `getref':
> refdbdref.c:4626: warning: passing arg 2 of `iconv' from incompatible
> pointer type
>
> ...I (hackingly, I'm sure) added in the following lines to refdbdref.c...
>
> #define ULONG_LONG_MAX (2ULL * LONG_LONG_MAX + 1)
> #define LONG_LONG_MAX __LONG_LONG_MAX__
>
> ...(I did not upgrade my libdbi drivers since using 0.95 - I am using
> libdbi-0.8.1 and libdbi-drivers-0.8.1)...
>
These are the newest versions anyway. refdbdref.c includes dbi/dbi.h,
which in turn includes limits.h, which in turn should include
sys/limits.h. Could you please check whether the latter indeed does
not contain definitions for these values? Which OS is this anyway?
regards,
Markus
<entry type="CONF" id="4" citekey="Phadke1994">
<part>
<title type="full">Application of global positioning systems to
electrical systems synchronized phasor measurements techniques and uses</title>
<author>
<lastname>Phadke</lastname>
<firstname>Arun</firstname>
<middlename>G</middlename>
</author>
</part>
<publication>
<title type="full">Colloquium on Developments in the Use of Global
Positioning Systems (GPS) in Power Systems,Feb 8 1994</title>
<pubinfo>
<pubdate type="primary">
<date><year>1994</year><month>02</month><day>08</day></date>
</pubdate>
<issue>032</issue>
<startpage>4/1</startpage>
<endpage>4/2</endpage>
<city>London,USA</city>
<publisher>IEE,Stevenage,Engl</publisher>
<serial>0963-3308</serial>
<address>Virginia Tech, Blacksburg, VA, USA</address>
</pubinfo>
</publication>
<set>
<title type="full">IEE Colloquium (Digest)</title>
</set>
<libinfo user="markus">
<reprint status="NOTINFILE"/>
</libinfo>
<contents>
<abstract>GPS navigational satellites provide an ideal method for
distributing a common timing pulse at any point in the power network. Receivers
of GPS transmission can reproduce the 1 pulse-per-second at any location on
earth with better than 1 μsec accuracy. Thus, if GPS receivers are built
into each measuring system, phasors measured by these systems could be used as
simultaneous measurements from all points of the power system, offering far
more accuracy than needed for performing engineering analysis
functions.</abstract>
<keyword>Global positioning system</keyword>
<keyword>Electric power systems</keyword>
<keyword>Digital computers</keyword>
<keyword>Electric current measurement</keyword>
<keyword>Voltage measurement</keyword>
<keyword>Communication satellites</keyword>
<keyword>Satellite relay systems</keyword>
<keyword>Iterative methods</keyword>
<keyword>Algorithms</keyword>
<keyword>Units of measurement</keyword>
</contents>
</entry>
backend-risx.c.tar.gz
Description: backend-risx.c replacement
--
Markus Hoenicka
markus.hoenicka@xxxxxxx
(Spam-protected email: replace the quadrupeds with "mhoenicka")
http://www.mhoenicka.de
|
