Please use this way to translate other strings like that. A callable
lambda for a _default wouldn't evaluate right, we do have to define
a proper function for _() to work.
bzr revid: p_christ@hol.gr-20101209100053-fjkoxjayaiqf8vj0
Yes, I have been subclassing the DAV module and improving on the existing
code, regarding that we couldn't ship our own version of the whole
python-webdav library (as in v5.x).
At commit b2a2c3f95416788, on 2010-04-06, we explicitly deleted our custom
copies of the DAV module and linked to the upstream project, in fact.
bzr revid: p_christ@hol.gr-20101209095950-qf2ppgdxw8y2lake
This replaces the pseydo-locking of python-webdav library with a real
db-based locking. Locks are stored as DAV properties, which will
effectively also be listed in the PROPFIND response of the nodes.
With locking in place, Office suites can collaborate on documents online.
bzr revid: p_christ@hol.gr-20101207134041-8negkvxrbscv7fs7
If you are trying to expose your OpenERP server through a DNAT box,
the propfind (aka. folders listing in WebDAV) must reference the external
IP/host that the client asked for. Use the more expensive "get_baseuri()"
of the interface class for that.
bzr revid: p_christ@hol.gr-20101207134005-v3oxr7zwbheiylxv
Similar to the dir.property model, ir.attachments may have arbitrary
properties, too. Note that we hide them from normal users, only list
them in separate form (don't want to bloat the ir.attachment one, even
at the XML level).
bzr revid: p_christ@hol.gr-20101207133955-8n6ipccdnygnuncn
During a database upgrade from cmdline, the HTTP server will be inactive,
so it would cause a deadlock at the WebDAV tests. Having the HTTP port
answer requests before the db is ready would be a bad idea, so we'd better
just timeout and skip the tests.
bzr revid: p_christ@hol.gr-20101207133947-bbni3x48wo3yffa2
The idea is to keep all current Latin alphabet translations in the new SerbianLatin language before changing the default Serbian translations to Cyrillic alphabet
bzr revid: odo@openerp.com-20101206093208-jiqnceg5zx7xvzjk
After some discussions, we decided that the desired behavior of the
ORM for one2many and many2many fields should be:
(foobar_id,'in', []) => empty result, because the list is empty
(foobar_id,'in',False) => find records that don't have a foobar
bzr revid: p_christ@hol.gr-20101129183413-yi1jzdwr1tis7d8t
The view_id had not been functioning as expected, and fixed today. So,
all these <act_window/> records that had it are now evaluated. These ones
broke after the server fix.
bzr revid: p_christ@hol.gr-20101127201656-bek6n0h4bxashwl2
It has been importing md5, hashlib, warnings, netsvc and tools, which
are not really appropriate for testing code. Instead, port to the
"test_reports" testing helper.
The rest of the file needs to be ported, too. TODO.
bzr revid: p_christ@hol.gr-20101127201630-zaaowb4oaoxvxlv1
This module had been inheriting stock.picking, and redefining the
"partner_id" column. Moreover, it has been redefining it with store=False,
which broke the parent stock.picking.
That's redundant, skip the whole class.
bzr revid: p_christ@hol.gr-20101127201619-qjmxvnw12kdqd2q2
Multiple coding malpractices (I can't name them errors, yet) were in those
lines:
- they used the netsvc.Logger() interface, which is deprecated
- they wrapped too much (we're not COBOL)
- they used capital letters for an object
- they defined a logger object that would live throughout the module
life, although it would only be useful for startup
- they used the gettext function, although there is absolutely no context
around to provide a language.
bzr revid: p_christ@hol.gr-20101127201408-gnn6js103fbpw8ou
P. Christeas