dbs [Tue, 4 Dec 2007 08:17:35 +0000 (08:17 +0000)]
Enable translation of seed values stored in database.
Moves all DML from schema files into a single file.
db-seed-i18n.py will generate POT files and SQL for new localizations.
Commit a new POT file for the database seed values.
miker [Sun, 2 Dec 2007 02:27:33 +0000 (02:27 +0000)]
This is a big commit. It started with a patch from Scott McKellar to
clean up several memory leaks in cstore and provide const-ification. I
was then inspired to add HAVING and multiple predicates per column support.
From Scott:
1. I applied the static qualifier to a number of functions, and a few
variables, that aren't referenced by other files. I left as global
those functions that are registered for dynamic calls.
2. I found a number of cases where a call to buffer_data() was being
passed as a parameter to a function, guaranteeing a memory leak. I
added code to capture the pointer returned by buffer_data() so that I
could free it later.
3. I added code to free a number of growing_buffers, jsonObjects,
and dynamically allocated buffers that were otherwise leaking. In
some cases I had to juggle some extra pointers so that I could
distinguish between things I could delete and things I couldn't. In
other cases I rearranged operations so that the freeing wouldn't get
bypassed. I don't guarantee that I have plugged all the leaks.
4. Whenever possible I replaced buffer_data() with buffer_release(),
to eliminate a round of malloc() and free(). In some cases this
change enabled me to eliminate an intermediate variable.
5. Whenever possible I replaced jsonObjectGetKey() with
jsonObjectGetKeyConst(). Sometimes I had to juggle an extra pointer
or two in order to distinguish between what was const and what wasn't.
6. I applied the const qualifier to a number of local variables and
function parameters.
7. Here and there I changed double leading underscores to single
leading underscores.
dbs [Sat, 1 Dec 2007 19:39:57 +0000 (19:39 +0000)]
Use XML::LibXML's parentNode function to avoid invalid expression warning.
Short-circuit test for driver -> language correspondence to avoid bogus warning noted by Ben Ostrowsky.
miker [Mon, 26 Nov 2007 18:25:09 +0000 (18:25 +0000)]
Big cleanup of the C auth app from Scott McKellar:
1. Apply the const qualifier where possible.
2. Plug memory leaks.
3. Make some global things static.
4. Some minor tidying-up.
Details:
1. The variables __oilsAuthOPACTimeout, __oilsAuthStaffTimeout and
and __oilsAuthOverrideTimeout are now static, since no other file
references them. Also I reduced the leading underscores to one each.
2. In several places I replaced jsonObjectGetKey() with
jsonObjectGetKeyConst().
3. In oilsAuthInit(): I plugged a memory leak by moving the freeing
of username out of an else block and into the code following.
4. In oilsAuthVerifyPassword(): I plugged a memory leak in the case
of an early return.
5. In oilsAuthVerifyWorkstation: I plugged a memory leak by freeing
workstation at the end.
6. In oilsAuthCheckCard(): I eliminated the ctx and userObj parameters.
The function didn't do anything with them except to verify that they
weren't NULL. I transferred that responsibility to the calling code
in oilsAuthComplete() (where I also plugged another memory leak).
7. Also in oilsAuthCheckCard(): I plugged a memory leak by freeing
params and card. Also I rearranged the last few lines a bit so that
we free active in only one place.
8. In oilsAuthComplete(): I plugged a number of memory leaks caused
by a failure to free userObj in various early returns.
9. Also in oilsAuthComplete(): In order to be able make uname a const
pointer, but still be able to delete it when necessary, I created an
intermdiate non-const pointer named freeable_uname -- which also takes
the place of the original freeuname variable used as a boolean.
10. In oilsAuthSessionRetrieve(): I moved the freeing of evt out of
the if-else structure and put it at the end, to make sure that none
of the branches would miss it.
11. Also in oilsAuthSessionRetrieve(): one of the branches of the
if-else had a second variable named evt, hiding the variable of the
same name declared in an enclosing scope. That's not a bug, but it's
confusing. I renamed it to evt2.
dbs [Tue, 20 Nov 2007 00:56:49 +0000 (00:56 +0000)]
Add current POT (PO template) files for translators.
Change "updatemoz" target to "updateproject" anticipating non-Mozilla PO files.
Remove duplicated English entries from French translation of opac.dtd.
Remove en-CA and fr-CA opac.dtd files; these will be generated in future releases.
erickson [Fri, 16 Nov 2007 16:44:59 +0000 (16:44 +0000)]
Storing the compiled search blob in a cookie. That cookie is used to populate
the advanced search widgets. (This means we don't have to re-parse the query syntax
in the OPAC).
Removed the hold cancel option from the edit column since Cancel is an
option in the new hold actions dropdown.
erickson [Thu, 15 Nov 2007 22:01:45 +0000 (22:01 +0000)]
capturing any preceding search terms and appending them to the default search class. rolling back previous change which allowed the query string to complement existing searches in the hash.. now the query string is the only source of class-based search data.
dbs [Wed, 14 Nov 2007 16:18:20 +0000 (16:18 +0000)]
Add a basic 'make test' root Makefile target.
The initial test suite validates the example XML files where a schema is available
or checks files for well-formedness where a schema is not available.
dbs [Sat, 10 Nov 2007 02:06:00 +0000 (02:06 +0000)]
Delete reporter cruft.
Rename success and failure templates to match config file defaults.
Ensure current reporter code gets installed as part of openils_all target.
erickson [Thu, 8 Nov 2007 23:58:14 +0000 (23:58 +0000)]
Made sort labels more explicit when an option has already been chosen
added relevance as a sort option -- probably needs some kind of style change
since it doesn't match the style of the other top-level labels
erickson [Thu, 8 Nov 2007 15:37:27 +0000 (15:37 +0000)]
Committing initial limit-to-available and sort options on record result page
Also, cleaned up some old code and did some tab cleanup
TODO:
Automatically re-fire search when limit or sort options are changed
Add relevance as a sort option
Add limit to advanced search
Make limit and sort sticky on advanced search
Evaluate dropdown labels - i.e. What is clearest to the user in the smallest amount of space?
dbs [Thu, 8 Nov 2007 04:01:27 +0000 (04:01 +0000)]
Low-hanging fruit for better (X)HTML compliance:
* Add type and language attributes to all <script> tags
* Close empty tags (use <img /> instead of <img> </img>, etc)
* Use Unicode character entities for next/previous
* Place some elements within legal parent element
miker [Tue, 6 Nov 2007 14:22:39 +0000 (14:22 +0000)]
adding internationalization core table to the IDL; adding I18N flag to the obvious first round of translateable fields via fm_IDL.xml; remove debugging from the i18n stored proc
projects / Evergreen.git / log