IF YOU WOULD LIKE TO GET AN ACCOUNT, please write an
email to Administrator. User accounts are meant only to access repo
and report issues and/or generate pull requests.
This is a purpose-specific Git hosting for
BaseALT
projects. Thank you for your understanding!
Только зарегистрированные пользователи имеют доступ к сервису!
Для получения аккаунта, обратитесь к администратору.
height of 100% means that if the enclosing widget is 100 pixels high, this
widget should be 100 pixels high. If this widget is then placed someplace
other than at the top of the enclosing widget, e.g. with top:20, the bottom
of this widget will fall off the bottom of the enclosing widget.
The normal way to solve this is to specify top and bottom attributes rather
than top and height attributes. Also, since widgets often look better if
they don't extend all the way to the very bottom, specifying a bottom
attribute equal to the top attribute typically has an appealing appearance.
I've fixed the tree widget so it no longer falls off the bottom of the page.
(This used to be commit 44d5f5ec6d)
appear. Now that the finite state machine properly ignores events from
widgets which are not handled, the initialization of the tree should be done
when the module's canvas appears (i.e. just after the canvas has had the
module's gui applied to it).
(This used to be commit ab620c9554)
impossible to add object without an objectclass), we need to
use the default ldb search filter (an empty one), to retrieve
all objects, and all record contents.
(This used to be commit fab315bb8c)
but ...
- we have a problem with removeAll() in ldbmodify
- we seem to not properly cactch errors, we always return a "success"
alert even if the operation is not successful :(
(This used to be commit 8ed568ae10)
or needs scroll bars; rather, it's internal areas within the tree widget
that provide the scroll bars. (The next version of TreeVirtual will prevent
setting overflow to anything other than "hidden".)
(This used to be commit 2dc0748640)
allows the Ldb Browser tree to properly size itself upon initially appearing
and upon window resizes. There are still a few problems with it that I need
to resolve, including an occasional set of double scrollbars, and making it
resize the tree column when the splitter is resized.
(This used to be commit c3c93ad36a)
that Simo asked for:
* One can now open a branch without selecting the row. The previously
selected row (if any) remains selected.
* Selecting a row does not automatically open the branch. That can be done
by either clicking on the open/close button or pressing Enter while that
row is selected.
* The entire tree can now be disabled. In LdbBrowse, the fsm is configured
to disable the tree during each remote procedure call, so that a pile of
requests don't get queued during RPC.
The most obvious flaw with TreeVirtual right now, is that the tree does not
properly resize (or even initially size itself) to its container. Fixing
that is my next task. It's actually a generic Table issue that people have
been asking for for ages: having columns resize themselves automagically.
(This used to be commit 23a8b77763)
only those icons which are actually used. This means that instead of a few
thousand files getting copied to /usr/local/samba/share/webapps (or
wherever), the number is less than an order of magnitude fewer. This should
make Andrew happier yet.
(This used to be commit 754c4fe01f)
Still I don't see the cursor :(
Swap -/+ buttons, you risk to delete an attribute by mistake when instead you want to add a new one,
and you might even not notice it.
Simo.
(This used to be commit afa61d5971)
withstanding, although many web sites ignore Gecko and work properly only when
accessed via IE, we should strive to a higher standard.
This check-in makes the Samba4 Web Application Framework work in IE as well.
(This used to be commit fed131747a)
More testing revealed that this isn't yet bug-free... Sigh. One of these
days I'll start writing bug-free software...
- Status and Statistics timer wasn't always stopped properly when switching
away from that module's display. It seems silly to continue to update
status and statistics when the page isn't being viewed.
- Single instance of the Login window was not accessible if one switched
between modules.
- Upon return from a Session Timeout, we can retry the original RPC request,
but it may fail due to information that was stored in the old session. This
was the case in the LDB Browser module, where the open database handle
became gone with the session. We now detect a Resource Not Found and
re-open the database.
(This used to be commit 09a8292033)
- Add authentication. The Web Application Framework can now be called
directly and it will rqeuire authentication if required, and should re-query
the user to log in when the session expires.
- General clean-up
(This used to be commit 27c5d7dca6)
- Fix handling of predicates in finite state machine
- Make use of predicates to handle a specific PermissionDenied failure, vs
some other generic failure
- Add JSON-RPC error constants
(This used to be commit 38ae9658cb)
- Disallow, for now, any ScriptTransport access. A serious security issue has
been described, and since we don't currently need it for anything, disable
it completely.
- Continued clean-up towards implementing the common authentication code
(This used to be commit 07817a5489)
- correct an error message in the JSON-RPC server
- provide a common RPC request function
- make it easier to have multiple pending RPC requests. this allows an
application-specified request to return "not logged in"; authentication
information to be requested and sent; and the original request to be
reissued.
- in statistics module, if a time is provided as zero, leave it blank rather
than displaying 1 Jan 1970. this showed up in the last_used_at field in the
SMB status report.
- clean up makefile: remove no-longer-used hard-coded path and install target;
remove "chmod" and "nice" commands which were part of the original skeleton,
since that is not deemed appropriate for samba makefiles
(This used to be commit 339ee73ce2)
- No path given to web server means use old scripts (missed check-in)
- Build Web Application (qooxdoo stuff) from autogen.sh since python isn't
available or is differently-configured on various build-farm machines
- Don't continually check in generated files
- Properly identify SWAT "Statistics" module as "Status and Statistics"
(This used to be commit 9d89b7663a)
