This reverts commit cb11cbd2f5.
Actually not needed after 3d36cac7fc4808f708f03b66f5c099de440e4569
in OOjs UI, and would change the UI in unexpected way.
Bug: T103403
Change-Id: Ia30b168ea29d03aa76ad81d1f9894a67604fdc08
When loading VE by clicking Edit on a view url, this naturally
works (we temporarily append our query, and remove it when ready).
When loading VE from a permalink (e.g. Open in new tab) or any
other source linked from HTML pages, then we're not in control
over adding the query (it's already there); we're only in control
over restoring the view url afterwards.
Bug: T102363
Change-Id: I4912ff1c6b28ac987517760ffed481a4cd3bd1ca
All this is no longer needed since 121216184718e7ebe22d6ecc8f5af5fe4e202465 in
OOjs UI and started causing issues after bb9c9c4f6a2fca6aed0e671e904e469d57da5f34.
Bug: T98795
Change-Id: I6493d6b52b313aac521aee2b0cff1571ea63bbe5
* Add a 3rd stage will triggers on API load
* Allow stages to complete in different order (as happens
on second load)
* Use CSS transitions
* Start slow and speed up using ease-in and progressively
shorter transitions
Bug: T95137
Change-Id: Ib9090e6c80b7a186bf353c40921d40456d90211a
The toolbar.destroy() call that was added in 5c38995bd9 was
needed to destroy the dummy toolbar when tearing down VE
before a surface had been intialized. But when tearing down
VE after the surface had been initialized (e.g. exiting using
the Read tab), we would do:
* setTimeout in tearDownToolbar()
** Set toolbar height to 0 and wait for transitionend
*** toolbar.destroy() once transition complete
* toolbar.destroy() in cancel()
This meant that we'd listen for a transitionend event
on an unattached element (because once the setTimeout runs,
the toolbar has already been destroyed by the second call),
which of course never fires, so we'd never resolve the
tearDownToolbar deferred and never finish tearing down VE.
Bug: T98388
Change-Id: I504f0cb0bf13643773fc98cb18b7b380cccb2f88
This is no longer needed, as VisualEditor now loads the content
from the /api/rest_v1/ entry point at the main project domain.
This reverts commit c741db5378.
Bug: T97500
Change-Id: I9b0dba58484884cdf3c7917c633b9815de4c9f92
We need it for the strip-if-empty logic in MWHeading,
and for consistency with desktop in general.
Bug: T96395
Change-Id: I9e70df896417811df087f58f3107919ba704f7c4
Easy-Deflate.js and its dependency, Base64, are not needed for editor
activation. Defer loading these modules until the editor has been activated.
Also promisify prepareCacheKey.
Task: T94616
Change-Id: I2e754fc835a5608b27d81117e1fbc9ea97d5744b
The activation timing was always a bit of a lie even pre-TargetLoader,
because the timer only started when the first RL request for VE
modules had loaded. But at least the process it covered was consistent,
which is no longer true with TargetLoader. Now that we start the
request for the HTML together with the RL request, the activation
time might include some, all or none of the HTML request depending
on how fast the RL request was.
This change makes the activation timings more useful by measuring
from the moment the user clicks "edit" to the moment the editor
is done loading, which is what actually matters.
* Moved start of activation timing to VPT init
** For mobile this falls back to when mw.Target#load is called;
we'll have to fix that in MobileFrontend later
* Moved end of activation timing out of TargetEvents#onSurfaceReady
into individual onSurfaceReady handlers
** This is necessary because VPT's onSurfaceReady does quite a lot,
and we want to include the time that takes in our measurements
Change-Id: Ie44f0b839b39a2b3b22dcd86e20f0d1170cb6069
/_preconnect is a special end-point that is handled by Varnish with an HTTP
204, sparing the RESTBase backend. See I95a716592.
Change-Id: I0c0430014768d7a1c6673d078569d0cf4062d338
Keep VisualEditor load times snappy by eagerly establishing a connection with
RESTBase via a beacon request, which is deferred until after window.onload /
setTimeout to ensure it does not slow down the loading of the page.
Task: T94784
Change-Id: I19fd2ef6beffe1c4c4f05c2716da079b49e88b95
If70ff601 didn't really fix anything. The only reason you'd want this data is
to distinguish between mobile and desktop data, but it just set the value to
the desktop version always...
Bug: T95432
Change-Id: I76722e3ad8b7dbe644374b24093bec696f27f48c
When deactivating, verify first that the welcome dialog was
initialized before calling 'close' on it.
Change-Id: Ife98b396f3d8641e2bd313c549fe867bcd84db9b
Split up beta and meta dialog show methods so that beta dialog
is displayed as soon as possible, regardless of the surface being
ready. Also make sure that we destroy the temporary window
manager on destroy.
Bug: T90454
Change-Id: Ib8f94518af431487ce940a74a8c268dbdbe403d2
Due to changes in the way VisualEditor is loaded, the trace.activate.{enter,exit}
events no longer enclose the entire VisualEditor initialization process. This
change ensures that trace.activate.enter is emitted as soon as the user clicks
on an edit link / tab and that the trace.activate.exit is not emitted until the
toolbar has been fully activated.
Change-Id: Ief798faa95a58898b9ae4dcbbbd30506c5dbd9a7
On short pages which don't extend below the fold, the progress
bar should center within the #content element.
Change-Id: I0b99e42f5bab8177d3b4ca6dd372d6403ae9b136
Relies on I69cf0a88
Using our own new message because TitleBlacklist's own one includes text that
we can't parse on the client and relies on a parameter we don't get from the
API.
This relies on WikimediaMessages' version of the Edit schema being updated at
the same time.
Change-Id: I4c75369b8b97973b72899bfaecbd5a996a440c68
Set 've-activated' as soon as edit is clicked, with a 've-loading' state.
This necessitates moving the relevant styles to mw.ViewPageTarget.init.css.
Change-Id: Ic9757cdbf63a2f72eda0dd03ff5588d79028ba0e
Move requestPageData from mw.Target to TargetLoader, call it
in init init, and pass the promise it returns into load()
via activate().
Bug: T90372
Change-Id: I828b8474e5a76b3d0d7d08735b4d865c29d2f820
This introduces TargetLoader, which manages plugins and RL modules
in a slightly more generic fashion so that Targets themselves don't
have to. This allows us to load all RL modules in one load.php
request, rather than first loading ViewPageTarget which then
loads the other modules.
TargetLoader loads in the bottom queue, so it will be loaded
as part of the main load.php request, but in VPT.init.js we
still have to wait for it with using() because it might not
have arrived yet. This also degrades gracefully on cached pages
where TargetLoader isn't in the bottom queue: it'll be loaded
as a separate request instead, which is suboptimal but no
worse that what we were doing before.
Right now TargetLoader is small enough that it could also be in
the top queue, but in the future we want to add things like
the action=visualeditor API request to it, and mw.Api is
relatively big.
Note: this also makes a breaking change to the plugin API:
plugin callbacks no longer receive the target instance
as a parameter, as they're now executed before the target
has been constructed rather than after. In the long term,
if we want to give plugins access to the target instance,
we could give them the target promise somehow. For now,
I've killed this feature because nothing used it and
the change from a direct object reference to a promise
would have been a breaking change anyway.
Also fixed incorrect documentation index for ve.init.mw.ViewPageTarget.init.
Bug: T53569
Change-Id: Ibfa6abbeaf872ae2aadc6ed9d5beba7473ea441a
Move the surface focus() call to be done after VE has loaded, to
make sure that the position of the content editable field is at the
top of the page and not -- as happens in Firefox -- after the read
page contents.
Bug: T90420
Change-Id: If91cea42c083d67b1ee6396402c3f607dde70471
Depends on Ib9471bc0 in VisualEditor core. Without that patch
this is actually necessary, because we were removing some event
handlers in the meantime.
Change-Id: I145f1891efd1c91eeb6154e11e96e3fc160e2be3
We used to attempt to not break Firefox's bfcache, but this
didn't really work very well, and it's not clear that avoiding
bfcache breakage is even a good idea. It's more sensible and
consistent to deliberately break bfcache while VE is active.
On top of that, the Firefox people believe that our trick
shouldn't even have worked to begin with:
https://bugzilla.mozilla.org/show_bug.cgi?id=1102664
Because the onbeforeunload handler is removed when VE
is deactivated, bfcache still works if you first click
Read, then navigate away.
Also clean up the management of the unload handler, using
addEventListener() and removing the return value fallback
code that is needed for beforeunload but not unload.
For beforeunload this is harder to clean up because
the addEventListener()-based API for returning a value
isn't consistent across browsers.
Change-Id: Ie4fe9ea3a59a54ba462733aa5e42bfc0ed5b15eb
Still misses some preinit aborts because we need to figure out
a way to attach the unload handler early enough to catch these.
Change-Id: I0ce721e24e69c31318064c6b443c1bfe01077546
The toolbar now uses a CSS transition for its height instead of
JavaScript animation through slideDown().
* The animation is on toolbar instead of toolbar-bar so that it
contains the padding and borders. Otherwise it slides up until
there is the top and left (quite thick) borders stacked on
each other which then disappear at once.
* Add transform/translateY so that toolbar also transitions
when it is in floating state.
* Move class addition to attachToolbar() to avoid additional
reflow.
Bug: T89543
Change-Id: I30a7b69b77b874d220f60ebe7f7e616cd77bcc36
Still to load at this point are
* Additional modules (first load only)
* Parsoid HTML
To account for this we disable the toolFactory event listeners
to prevent flickering, and create a hidden blank dummy surface to
attach the toolbar to.
Bug: T76523
Change-Id: Iab24858f23f4db944dcaa6683a82b950ea9ee1b1
We can't use wgTitle directly because it has spaces
but no namespace prefix (e.g. "Foo bar"), and we can't
use wgPageName directly because it has the namespace prefix
but underscores instead of spaces (e.g. "Project:Foo_bar"),
so run wgPageName through mw.Title.
Bug: T90433
Change-Id: If431f5b9f6efa2690cfd964f80d6ac71cb568087
Follows-up 62c1d64ad0.
* Remove '.mw-indicators' animation for opacity. Was a left-over from other
elements previously stripped in 62c1d64ad0.
* '#siteNotice' no longer slides. It's hidden at the same time as '#contentSub',
'#toc', and other elements. This is an annoying jump, and we should find a
way to do it consistently and in a performant manner.
* Fix state classes added in 62c1d64ad0. During deactivation, the ve-active
class was kept for too long. This created a temporary state in which ve-active
*and* ve-deactivating were applied. This was harmless as nothing ises the
've-deactivating' class yet.
* Leaving toolbar as-is for another commit.
Bug: T89543
Change-Id: Ide49f93c45a12f2c13eb4412a9a16a82fefeb43e
Follows-up I002322beaae. I had this in the local patch but got
lost in rebasing. The missing return statement caused an uncaught
error when clicking section edit links as it was not forwarding
the return value.
Change-Id: I66b6a789f7182a4446122e94f19cfdb324f8d675
De-duplicate the logic of:
* Call showLoading()
* Call getTarget()
* Call activate()
* Bind always(hideLoading)
Also use then() instead of done() on getTarget() so that failures
propagate to the always() handler. Previously the interface would
indefinitely be in a loading state if target fails to load.
Change-Id: I002322beaae64c0de96457eb56dbc68a5fc16369
Tracked stages are:
* activate: wraps the entire activation
* domLoad: API request for Parsoid HTML
* parseResponse: parse HTML, interpret things like link cache data
* convertModelFromDom: create linear model from Parsoid DOM
* createSurface: build DM and CE trees, create DM, CE and UI surfaces
* initializeSurface: initialize UI and CE surfaces, and CE nodes
* setupToolbar: toolbar setup and attachment
* initializeToolbar: toolbar initialization (deferred so not wrapped by activate)
Bug: T89536
Change-Id: Ife5664e872bad68dcf3c123f274f9c8a38a06031
It's slow, especially on large pages, and it's triggered
very infrequently these days, and only for known bugs.
In the future we should replace this with a debugging
interface that displays the DOM diff between the original
DOM and the round-tripped DOM, as opposed to the boolean
interface we have now.
By extension, this also means the visualeditor-needcheck
tag won't be applied to new edits any more, although
its registration and messages are kept around because
edits with this tag still exist in page histories.
Bug: T87161
Change-Id: I909153492a5786b4b69fccd42ce3c1d4bdb3a059
Originally added in 5036099906.
* Account for bug T89513 for 'tempdisable' (was done for 'enable' in 9cba9910c6).
* MediaWiki core now loads user defaults in a separate module not
subject to page caching, making our workaround obsolete.
Keeping defaultthumbsize as that is intentionally ignoring the user's
preferences.
Change-Id: I07a7aa8e08c728ddeba67f36a98e989c220dcb50
* Ensure that the toolbar remains inside the target so that
target triggers still work
* To keep the page title beneath the toolbar, move it inside
the target on setup, and restore it on teardown
* Have teardownToolBar/DebugBar return promises so we know
when teardown is completely finished.
Change-Id: If2a9ce9a653aed51d21c8e8dd7795f2533d54f70
* Use .ve-activated for elements changed in activate()
- hideReadOnlyContent()
- mutePageContent()
- mutePageTitle
* Use .ve-active for elements changed in onSurfaceReady()
- hidePageContent()
* Set 've-activated' class from activate() instead of transformPage()
to consolidate reflows and minimise DOM interaction. It's still
in the same (synchronous) execution path, but a few statements
earlier now.
* Remove obsolete #toc wrapper. This <div> (with data property
to distinguish it from potentially foreign parents in the future)
was there to aid slideDown/slideUp animations, because those
don't work well on table elements. See eba7d58dd1.
* Remove obsolete setTimeout in restorePageTitle().
The removal of .ve-init-mw-viewPageTarget-pageTitle was delayed
by one second (introduced in Ibc3fa2fb7 / 4cc88b9850). This
was to account for a jQuery animation we no longer use.
* Remove unused '.ve-init-mw-viewPageTarget-transform-muted'
* Remove unused '.ve-init-mw-viewPageTarget-transform'
* Remove unused '.ve-init-mw-viewPageTarget-pageTitle'
The resulting stylesheet exposes that we're not consistent in
whether elements hide immediately (ve-activated) or once the
surface is ready (ve-active). This is intentionally kept as-is
within this commit.
Of the different elements that had their opacity changed, only
firstHeading was being animated. This animation was removed.
Bug: T88590
Bug: T87160
Change-Id: I87033456f715d99a88425e38e8ac5171144f4ec8
Its white overlay is pretty good at obscuring them, and the notices popup
would close when the dialog is closed, making it very easy to miss them.
Also fixed indentation.
Bug: T86987
Change-Id: I9f93c4169de98ef77a1eb33fd31a0ec524b6d2b7
MobileViewTarget.
Due to this being missed, pageName was getting set to a jQuery object,
which eventually got passed as a data value in a $.ajax call, causing
jQuery to emit a TypeError: Illegal invocation.
I guess this needs a MobileFrontend commit as well. Don't think it fixes
all the issues with loading VE mobile though.
Bug: T78710
Change-Id: I431a902563f608a7a33028cd60112f39c57000a4
Local changes:
* Call bindHandlers()/unbindHandlers() in activate/deactivate
* Extend onDocumentKeyDown rather than overwriting it
Change-Id: I0739e71781c5140c54512ab421bdcb6d1df5fbbe
Follows-up b0e8900, 447e8de, d3f26e65.
Contrary to binding a reference, this-overiddes for inline
functions gets messy and fragile. There's also a minor
performance overhead that can be avoided. No separation of
concerns, eiter, as the object is held by the closure.
Also:
* Remove left-over instances of @method.
* Remove redundant '0' argument to setTimeout.
* Use null instead of $ for apply() on static methods like $.when.
* Use .empty().append() instead of .html(). Does the same but
without additional overhead and without confusion association
with "html" which isn't used (it doesn't serialise and re-parse
or clone, it just appends when given one or more nodes).
Change-Id: I797a3667c67d52568150be9be5d043d149f22077
Various VE elements use escape to perform tasks. Particularly
difficult are OOUI dialogs which listen to the document and thus
fire after this target's listener, so defer and check if the
event was previously stopped.
Bug: T78201
Change-Id: Ie335f199766923833588202626c3af8e4cbcb9b5
New changes:
9929b6d [BREAKING CHANGE] Make ve.init.target an OO.ui.Element
Local changes:
* Fix target constructors now they generate this.$element
Change-Id: Ia6bf604c67305834a3f73863a10336cfbb83fa77
New changes:
776e801 Delimit text nodes in debug observations
72643bc Remove debug bar hiding now it is handled by the surface
cf1e167 [BREAKING CHANGE] Multiple surface support and demo
Local changes:
Account for detachable toolbars in core being setup in a different order.
Change-Id: I29156948711d55ba5f6dd4f6c35ca0e4cca32373
This doesn't quite tell us about the user's first interaction with
VE (if they open a dialog, for instance, it could be minutes before
that results in a transaction), but it's a good start.
Bug: T76524
Change-Id: I6bac987332802d7363ed65a471b2b1ece67086d8
New changes:
fc4af20 Fix filibuster expansion bug
3feadb2 [BREAKING CHANGE] Refactor for multiple surfaces
Local changes:
* Use target.getSurface now that targets can have multiple surfaces
* Remove debug bar code now surfaces manage their own
* Remove import rules code now implemented in base target
Change-Id: I18d6f9f2f36431959d31347b4c1952a1aca1350e
It's good that we're not trying to close the save dialog
if it's not open, but we still need to release our reference
to it in that case, otherwise trying to open the save dialog
in a new incarnation of the editor will fail.
If you opened VE, opened the save dialog, closed it,
went back to read mode, opened VE again, and tried to open the
save dialog again, it would fail because it tried to open
the cached reference to the previous editor's save dialog.
Bonus: prevent JS errors in this case by only attaching an
onClose handler if the dialog actually opened successfully.
If opening the dialog failed, the callback parameter isn't
a promise but an OO.ui.Error, and calling .always() causes
a JS error.
Change-Id: I62d6d3e7cd3df2c29a0c99e2e23bf1aac5187a12
New changes:
5eb7294 Find and replace
123a6ad Find and replace tool
557f70f Go backwards through find and replace results on shift enter
4a4ed81 Allow language widget to filter on available languages
Local changes to adjust.
Change-Id: I15e14d6d3c7ee60e465baf3a4df7572851cbca3e
Both onDocumentKeyDown and onViewTabClick would check
this.activating and invoke deactivate() differently depending
on its value. That's ugly, so incorporate the "don't show a dialog
if deactivating during activation" logic into deactivate() itself.
Rename the override parameter to noDialog, and fix the broken
logic around it (it overrode too much).
Also move setting this.activating = false; out of these event
handlers into cancel(), where it clearly belongs because
activatingDeferred is being rejected there.
Change-Id: I6c040c74d36f829128a175f00d4e712dd19d31a7
