scottmk [Mon, 10 Aug 2009 16:14:58 +0000 (16:14 +0000)]
When defining a SELECT list for a given class: an empty
array no longer results in a default SELECT list. You get a
warning message (but not an error) and no fields from that
class. For a default SELECT list, use "*" or null.
erickson [Sun, 9 Aug 2009 14:35:19 +0000 (14:35 +0000)]
Patch from Galen Charlton to add Version to application.ini
==
The attached patch restores the Version key to the staff client's application.ini. This key is now required; without it, xulrunner-bin --install-app on OS X will silently crash, leading to an incomplete and non-functional installation of the client on that platform.
Note that when the client is built, the value of Version is replaced by the build ID.
scottmk [Wed, 5 Aug 2009 18:58:20 +0000 (18:58 +0000)]
In oils_cstore.c, function searchWHERE(): rewrote a loop to use an
index instead of a jsonIterator, to traverse a jsonObject that is
known to be a JSON_ARRAY. This change avoids the malloc and
free required to create and destroy a jsonIterator.
scottmk [Mon, 3 Aug 2009 03:02:48 +0000 (03:02 +0000)]
In the initialization routines of oils_cstore.c: recoded a couple
of loops to use an osrfHashIterator, instead of creating an
osrfStringArray of keys and then traversing that. This change
eliminates over 500 mallocs and frees. In addition, traversing
the osrfHash with an iterator just hops along a linked list,
which is almost certainly faster than calculating a hash total
for every key.
When a JSON query refers to a column qualified by a table alias using
the plus-class trick (e.g. "+aou":"opac_visible"): verify that the
column belongs to the indicated class. If it doesn't, return an error.
1. Issue a warning if the SELECT list is empty for a given table alias
(other than the core class). This happens when we try to do the
equivalent of SELECT * for a non-core class.
2. Return an error if the generated SELECT list is empty. This can happen
in a case like the following:
don't fulfill hold when patron checks out a like copy if the hold in question has a copy on the holds shelf. Use the power of cstore to find appropriate holds
for now, disable refresh of entire patron interface after penalty manipulation. What we're really avoiding is the stop-sign page. Want to revamp how the patron interface works here to make use of observers.
Staff initials for standing penalties. Looks for org setting ui.circ.standing_penalty.require_initials to enable this feature.
The Apply/Modify button will jump to the initials textbox if empty.
No dedicated initials field in actor.user_standing_penalty, so appending to the note field. Not parsing notes for initials when editing, though we may want to (I'd argue for a dedicated field in
this case).
Reworking of the "Messages" interface in the patron display for manipulating standing penalties. Not yet complete, but functional.
TODO:
Need to add a field for staff initials if a specific OU setting is present, and need to add a section and some methods for "archived" messages. Also need to change how penalties are expressed in the summary sidebar. Also need thought on
how/whether/when this will replace the existing Notes infrastructure, since it can do the same thing but better.
NOTES:
Experimenting with xul toolbar/toolbox; there's a beep when you use the hot key for the Actions menu that this might be the cause of. Also, we're actually using the Dojo openils.PermaCrud code for
modifying existing penalties. Exciting :)
small net::server process to read MARC data from a raw socket and import the data into Evergreen. It was originally designed for OCLC Connexion imports, but could be used to import MARC from any source
open-ils.cat.biblio.records.merge uses in_db merge now to take advantage of URI handling (and speed, etc.). added a perm test to in_db_merge. removed old bib merge code
Thanks to miker_ and dojo 1.3.2, the <script> tag for loading dojo works now (if you configure djConfig in inline javascript rather than as an attribute on the script tag itself, and dojo.require() works if djConfig is given the baseUrl for dojo.js
added api call for calling the new actor.usr_delete stored procedure, which purges user data/xacts, anonymizes top-level user data, and transfers data to another user where necessary
Patch from Warren Layton for support of RIS format in SuperCat - thanks Warren!
===
I've put together another SuperCat feed, this type for the RIS format.
This can be used with most bibliographic citation software, such as
Reference Manager and Endnote. Combined with the previous "MARC text"
feed for RefWorks, most software should be covered.
I have only included the changes to add the feed and the XSLT file.
The patch (against trunk) doesn't include any changes to the OPAC, but
I'm willing to put something together if needed.
The MARC21slim2RIS.xsl transform makes a few assumptions:
* Any authors listed in 100, 110, or 111 fields are considered
"primary authors" (A1)
* Any individuals listed in 700 fields are considered primary authors
(A1) if their last name is found in the statement of responsibility
(245c). Otherwise, they are considered secondary authors (A2). It's
not 100% accurate, but it's certainly a step up from just turning all
700-level authors into A2s.
* Corporate authors listed in 710 fields are always considered
secondary authors (A2).
* Meetings in 711 fields are always converted into "Title Series" (T3)
(i.e., a talk at a conference would have the conference name listed as
the "Title Series", with the title of the talk listed as "T1", primary
title).
* Apart from field 520 (used for Abstracts, "N2"), no 5xx fields are used
* Subfields in the 650 and 651 fields are used as keywords ("KW")
Developer's Certificate of Origin 1.1
By making a contribution to this project, I certify that:
(a) The contribution was created in whole or in part by me and I
have the right to submit it under the open source license
indicated in the file; or
(b) The contribution is based upon previous work that, to the best
of my knowledge, is covered under an appropriate open source
license and I have the right under that license to submit that
work with modifications, whether created in whole or in part
by me, under the same open source license (unless I am
permitted to submit under a different license), as indicated
in the file; or
(c) The contribution was provided directly to me by some other
person who certified (a), (b) or (c) and I have not modified
it.
(d) I understand and agree that this project and the contribution
are public and that a record of the contribution (including all
personal information I submit with it, including my sign-off) is
maintained indefinitely and may be redistributed consistent with
this project or the open source license(s) involved.
This update to oils_cstore.c moves in the direction of supporting
table aliases in JSON queries (but we're not there yet).
We now maintain a stack to represent the top level query and any
subqueries nested within it. In each stack frame we store information
about the core class and any joined classes, including whatever aliases
are used for them. This stack enables us to look up the table
corresponding to a given alias.
Such searches now replace the old is_joined function for determining
whether a given class is in scope for a column in the SELECT or
ORDER BY clause.
In passing:
1. Rewrote the first few lines of searchJOIN() so that it wouldn't
be so convoluted.
2. We now detect attempts to do autojoins, or to join to the same
table multiple ways. These are the sorts of queries that need
table aliases to disambiguate the table references, The old code
would create invalid SQL and then let PostgreSQL complain about it,
Now we complain about it up front.
3. Corrected spelling errors in a couple of messages.
use the editor's allowed() method for checking perms. added context org to perm check. removed info/activity logs since editor logs queries at info and updates at activity already. changed 'record' param to 'record_id' for clarity. returning the new id on create instead of the editor object.
Add a few more tests, including one of a combined publication
date in the $yp pattern rather than in the $yc pattern (example
taken from LoC documentation).
specific cookie path not playing nicely when communicating with staff client cookies, in particular, change operator functionality. go back to global cookies
In oils_auth.c: changed the treatment of inactive accounts.
Previously, if someone tried to log on to an inactive account,
the error message reported to the client would identify the
account as inactive, without regard to the password.
Now the message identifies the account as inactive only if the
password matches. Otherwise it reports it simply as a failure.
Also: changed the barcode to a pointer to const, pointing to
an existing string, rather than allocating a separate copy
that we have to free later.
Also: tinkered with a couple of info messages to avoid
invoking undefined behavior when barcode is NULL.
If you try to log in with a non-existent username, the old
code would report that the username was inactive. Now it
reports a login failure. Inactive usernames are still
reported as inactive.
added option to import holdings from main record import screen. Give the user the option to choose which item import profile, when bibs are done importing, attempt to import holdings for successfully imported records. Still needs a UI for configuring new import profiles and ways to import holdings from the queue interface
projects / working / Evergreen.git / log