Our spk currently have an issue with the bundled MongoDB (forked)
version that breaks if some files are inserted in a GridFS collection.
We need to find out a way to upgrade and migrate the bundled database
until we can re-enable the attachment feature.
Note that the feature is unmodified, it's just hidden in the UI.
This new draft saving system is currently only implemented for the
card description and comment. We need better a component
inheritance/composition model to support this for all editable fields.
Fixes#186
It seems that the total size of the sandstorm package manifest can't
exceed a certain size. Putting a small image works until I figure out
what the issue is.
* Grow the click-able zone of the list menu
* Remove a dead link on the list menu
* Merge list menu files with header menu to be consistent with the
board components internal organization
Closes#106
* Automatically display the overlay when the card details is opened
(previously we waited for the mouse to enter the card details panel)
* Improve the design of the minicards badges
* Change the minicard background when it is hovered or selected
* Removes unimplemented features links from the UI
* Fix the board canvas position when the sidebar is open (was hidden
behind)
Fixes#215
The new user interface uses the flexbox box model has it's primary way
to organize the layout. Unfortunately IE have a non-negligible amount
of subtitle and tricky bugs with its flexbox implementation.
This one was about a `flex-grow: 1` element not growing in a container
which size is defined with `min-height`. See the bug in Microsoft bug
tracker for more details:
https://connect.microsoft.com/IE/feedback/details/802625/Fixes#225
I decided to create my own and not to use a community package, because
1. it's straightforward
2. it's better integrated with our others libs such as EscapeActions
3. monitoring third-party packages evolutions (eg, CSS changes) is a
lot of work.
This is basically the same rationale than for our other generic UI
components such as the Popup/Popover.
This commit also slightly modify the general layout to remove
unnecessary wrapper DOM nodes.
This is an experiment about the implications of ES6 transpilation in
our code base.
We should also define a new ES6 style guide and a jsHint
configuration, for instance semi-colons are automatically inserted at
the end of lines, so we may remove them. We also need to figure which
ES6 features we want to use, currently I have followed Meteor-core
guidance which is reasonable.
Bower build process was sometimes unpredictable and had unspecified
dependencies (git) that caused bugs (eg #164, #172). Meteor packages
have better integration with the meteor build system and ensure
deterministic build, the downside being that we have to maintain
wrapper of JS package just for the purpose of publishing in
atmosphere. Others are also struggling with this, see for instance:
https://github.com/MeteorCommunity/discussions/issues/14
Hopefully we'll have a better solution one day (official browserify
integration? Who knows.)
As a side effect the new release of perfect-scrollbar breaks the
style, this will be fixed in a later commit.
Fixes#172
The new version of meteor speeds up the reload cycle, which is super
valuable during the development. I also removed the "imply-everything"
"meteor-platform" package in favor of a more fined-grained package
selection.
This version also introduces ES6 support with transparent babeljs
transpilation. Most features are enable (with the notable exception of
ES6 modules) and this commit started to use them in places where a XXX
comment suggested it.
Both the previous and the new value are completely arbitrary but some
users suggested that we reject names like "Li" for no good reason,
especially on private boards, cf #202.
Hopefully one day we'll have an administrator configuration panel.
Fixes#202Closes#210
Motivations:
* Iron-Router foces us to use Tracker.nonreactive black magic in order
to avoid un-necessary re-renders;
* There is a community consensus (supported by some MDG members) that
the flow-router API is easier to reason about;
* The useraccounts now supports flow router (that was a blocking
element when I considered the switch ~3months ago)
On the server we use the Picker router, as encouraged by the Kadira
team (which develop both Flow and Picker routers).
In the current state of things there are some bugs related to the
missing Loading architecure. Previously onRendered callback where
always called when the data the component needed was available, now
we have to handle this ourselves, which we will in a following commit.
This includes a security fix in mquandalle:bower, and a new package
for useraccounts iron-router templates which forced us to move some
configuration function calls.
