lp:~kamstrup/dee/bug-905463

Branch merges

Propose for merging

No branches dependent on this one.

Merged into lp:dee at revision 333

Michal Hruby (community): Approve on 2012-01-06

Diff: 11 lines (+1/-0)

1 file modified

dee/dee-shared-model.c (+1/-0)

api

Related source package recipes

No recipes using this branch. (?)

Create packaging recipe

Related snap packages

Related bugs

Bug #905463: Coverity FORWARD_NULL - CID 10029

Medium

Fix Released

Link a bug report

Related blueprints

327. By Mikkel Kamstrup Erlandsen on 2012-01-02

Fix bug 905463 reported by coverity. Possible crasher in DeeSharedModel when constructed without a swarm name

326. By Mikkel Kamstrup Erlandsen on 2012-01-02

Optimize the getters in DeeSequenceModel by shaving off a g_variant_ref() compared to the default impl in DeeSerializableModel. Benchmarks indicate a ~13% speedup when reading strings.

This branch also adds a benchmark for reading a string column out of a seqmodel.

It is now possible to pass arguments to tests/test-benchmark that specify prefixes that the benchmark names must match in order to be run. Without args all benchmarks are run.. Fixes: https://bugs.launchpad.net/bugs/900629. Appoved by Michal Hruby.

325. By Michal Hruby on 2012-01-02

Adds 3 new test cases for DeeTransactions (by mhr3) and fixes the bugs these tests turned up (by kamstrup).. Fixes: . Appoved by Michal Hruby.

324. By Michal Hruby on 2011-12-19

Fixes the order of parameters for dee_filter_new, updates vapi, integrates vapi generation into build system. Fixes: . Appoved by Mikkel Kamstrup Erlandsen.

323. By Michal Hruby on 2011-12-19

This branch implements possibility to create private connections which can be used to synchronize the shared models.. Fixes: https://bugs.launchpad.net/bugs/904299. Appoved by Mikkel Kamstrup Erlandsen.

322. By Mikkel Kamstrup Erlandsen on 2011-12-15

Add a human readable "Accuracy" field to the benchmarks to gauge the quality of the results. It's a pretty simple heuristic but should be statistically good enough for our needs.

Also adds two new benchmarks for DeeFilterModel using collators to sort the models by a column containing a randomized string.. Fixes: . Appoved by Michal Hruby.

321. By Mikkel Kamstrup Erlandsen on 2011-12-15

Implement transaction logic for DeeModels in DeeTransaction

This branch implements a new class, DeeTransaction, that implements
transaction isolation on top of any old DeeModel implementation. It
has only one interesting API call so far,namely commit() which does
as labeled.

How/if exactly we want to expose a transactional API on DeeModel itself
is still up in the air; but this work should provide a solid foundation
for that. (and all of the trickier code needed, anything else will be
mostly just gloss).

Since this is a rather complicated afair I recomment that we take a somewhat
functional approach to the review. Validate the testing harness - and possibly
submit a few new test cases to cover the stuff I didn't think of.. Fixes: https://bugs.launchpad.net/bugs/894023. Appoved by Michal Hruby.

320. By Mikkel Kamstrup Erlandsen on 2011-12-15

Overhaul DeeFilter and DeeFilterModel to work with GI and be more like DeeModelReader and DeeIndex

This branch gives a slight API break in how DeeFilters are created and reverses the order of the arguments to dee_filter_model_new() to match the order in dee_{hash,tree}_index_new(). I tested manually that Python works with DeeFilterModel now.

(resubmitted because of quantum entanglement with the no-const branch). Fixes: https://bugs.launchpad.net/bugs/904293. Appoved by Michal Hruby.

319. By Mikkel Kamstrup Erlandsen on 2011-12-14

Remove G_GNUC_CONST for all dee_*_get_type() functions. It is outright wrong to put that on get_type() function and have just been dragged on since forever in fear of breaking ABI.

Also fixes some compilation warnings in the test suite. Fixes: . Appoved by Michal Hruby.

318. By Mikkel Kamstrup Erlandsen on 2011-12-13

Emit signal by id, not by name, in DeeSequenceModel for a small perf gain

The benchmark suite indicates a 0 > x > 1% perf gain from this - but non-zero!
I expect the numbers to be slightly better on lower end hardware, and also when
looking on the actual spend CPU time, which I haven't done.. Fixes: https://bugs.launchpad.net/bugs/901098. Appoved by Michal Hruby.