The plan is to change the outer …TemplateWidget (which contains
a list of template parameter checkboxes) into a SelectWidget.
But this requires the elements in the list to be a subclass of
OptionWidget.
Note this change does not have any effect, as of this patch. But
this makes the following patches smaller and easier to follow.
Additionally:
* The OptionWidget class is already a LabelElement. No need to
initialize this twice. This happens via the parent constructor
now.
* Remove CSS that is not needed any more after Idc5e048. This is
not a FieldLayout any more.
* Update some related code documentation.
Bug: T285323
Change-Id: I92e8fd2bbece9e6c55083cdfe6ed7ad16a64d688
The icons have a padding of 6px around the icon image itself. To
get to the required 16px/8px space the margin was adjusted
accordingly. Note that there's also a 2px padding around the menu.
Bug: T272482
Change-Id: I3df9f355dfd5c4e6366432555b96bf788e784280
This is what actually happens:
* We call `addParameter()`.
* This triggers an `add` event.
* This calls an `MWTemplateDialog.onAddParameter` event handler.
* This code doesn't check if a parameter already exists (because
it shouldn't). It detroys the page in the content pane on the
right and recreates it from scratch.
The only reason we do this is to focus the input field on the
right. This patch introduces a dedicated event to do this.
Bug: T288827
Change-Id: I47effe05427cfabfcf534920edee79521eaa033f
The history.replaceState call in onReviewModeButtonSelectSelect() (1)
ignores any changes to the document URL made by external tools such as
RevisionSlider, (2) replaces with an empty string the history entry
state object if it was set by external tools such as, again,
RevisionSlider. Both 1 and 2 result in a wrong URL ending up in the
address bar at some point. This patch addresses these problems, (1)
creating a new mw.Uri() object every time
onReviewModeButtonSelectSelect() is called, (2) keeping the current
history.state object.
The original variable storing the URI object is renamed to avoid
shadowing (optionally can be replaced with its value as it's used only
once).
Bug: T288636
Change-Id: Ieb97b561a6c076aa28aae231fe286ac4d1051bbd
I found this while working on T274551, which is all about the
definition of "empty".
In the old sidebar a parameter's name is dimmed (gray) as long as
the parameter's value is empty. This stops working entirely when
there is a default value.
My first impulse was "this is a bug". When there is a default
value, both the empty string and the default value (when the user
enters it exactly) typically trigger the same behavior: The
template uses the default value, just as if the user entered it.
But this code is correct because of the way it is used. Only
parameters that are "truly" empty should be visually marked as
such. The moment there is a default value it is either impossible
to change this back to an empty string – meaning the parameter
can never be truly empty – or the empty string is meaningful user
input.
Bug: T274551
Change-Id: I90657bfe83e56afd3942428c0dd8a47b444e39c9
This should not have any effect on how the thing looks and
behaves.
* All elements in the sidebar should be reachable with the tab
key, including disabled elements.
* Enter jumps to the corresponding paremeter in the content
area on the right. But enter doesn't change the checkbox
state.
* Space canges the checkbox.
The class will be renamed in the next patch.
Bug: T285323
Change-Id: Idc5e04828ece0ba77a65e4c839cd3ffccc3b6733
As discussed in Ia44da16. This change avoids possibly hundreds
of events (when a template does have hundreds of parameters),
and replaces them with a single one.
Bug: T288202
Change-Id: Ic819e8c93e872b653c238f396f1f327b6a8759d2
There is still a lot to do, but this implements some basic
behavior that was missing before.
* You can now use the tab key to navigate all checkboxes,
including disabled ones (required parameters).
* Enter key now works on both the checkbox as well as when
the entire line is highlighted. Enter forces the checkbox
to be checked and moves the focus to the content area.
* Mouse clicks now work on the entire line. Before, only the
text label was clickable.
Open issues (not to be resolved in this patch):
* Clicking the text label and the empty space after the text
label does different things. Probably shouldn't.
* Should a click on the label check the checkbox?
* Space key should probably not move the focus to the content
area.
* Focus rectangle is different on disabled rows. Is this ok?
* Background color when a line is focussed is missing.
Change-Id: I22ccd1bea92e4f098d4b25a9e38cddde5c103423
The checkbox is the first parameter in the parent constructor.
The parent is the FieldLayout class. The checkbox becomes the
this.fieldWidget in the parent class. Just use this instead of
storing a duplicate reference.
Bug: T274543
Change-Id: I4ae7d467334f88f2be93a62660145a025089401f
I get a scrollbar at the bottom of the sidebar. The reason is
that the container's width is 100% + 1px. The extra pixel is
from the border, which is not needed in this mode.
Bug: T274554
Change-Id: I4f749be6b9a7f89f9a7a195dc66c5c18253b1327
This check makes sure the user doesn't loose work when clicking
the back button. I would like to argue that neither of these
values is valuable enough to block the user with a confirmation
dialog:
* Literally nothing is lost when the input is empty.
* The auto-value is only temporarily lost, but will
automatically be restored when the user decides to add the
template back. The input field is pre-filled with this value.
* The default value doesn't need to be manually entered. It will
show up anyway when the parameter is missing.
There is a rare edge-case, but it is not really relevant in this
situation. Some templates allow to override a default value with
the empty string. This will be considered irrelevant by this
code. However, this was already happening before and doesn't
change with this patch.
The only edge-case where this patch makes a difference is if a
parameter is marked as required or suggested, is documented to
have a default value, _and_ the template allows to override it
with an empty string. But this combination is rather crazy, if
not bogus, and not worth considering here, I believe.
Bug: T274551
Change-Id: Ib176a82844335c3d4dd5b720d335ec28245e1637
This is really only about the methods name, but doesn't change
any behavior.
I realized we work with several different definitions of what
"empty" means. There are at least two significant definitions:
1. When a parameter's value is the empty string or identical
to the default value, the behavior of the template is the same.
It will use the default value just as if the user entered it.
The auto-value is a meaningful value in this scenario and can't
be considered equal to the empty string.
2. The context here is when the user presses the back button.
This will destroy all user input. But an auto-value is not user
input. It will appear again when the user realizes they made a
mistake. Nothing is lost.
Personally, I would not use the word "empty" to describe this
concept. Things like "containsUserProvidedValue",
"isCustomValue", "isMeaningfulValue", … come to mind. These are
all still a big vague. A "user provided" value can be identical
to the default or auto-value. "Custom" how? I went for
"containsValuableData" for now.
Bug: T274551
Change-Id: I2912a35556795c867a6b2396cbad291e947f0ed6
This is a direct follow-up to I6ebd020.
Steps to reproduce the bug:
* Make sure you have a template with a deprecated parameter.
The position doesn't matter.
* Add the template. The deprecated parameter is hidden.
* Add an undocumented parameter, e.g. "b". This is added to the
end, as it should.
* Add an undocumented parameter "a". This should appear before
"b", but doesn't. The reason is because the invisible
deprecated parameter is in the list that is used to calculate
the index, shifting it by 1 (or more when there are more
hidden parameters).
This patch includes a few closely related changes:
* We can loop the list of checkboxes directly instead of
indirectly via the list of parameter names.
* I made it so that an active filter only resets if it would
hide the new parameter. The original problem we had to solve
was that the new parameter would always be visible, even if
it doesn't match the filter. This awkward mismatch is still
guaranteed to not happen.
Bug: T274551
Change-Id: I1b0480ae836cc19b77b159d3fb30ff32e8c59df4
I came up with a new event to do this. This event is triggered
individually for each parameter. An alternative is a single
event that gets a list of visible parameters. Is this better?
What do you think?
Bug: T288202
Change-Id: Ia44da16917c28171a01aef0f1c613dcd5d3266ba
This is – for now – intentionally done in a way that can be
undone. This will still be helpful for debugging for a while.
But we need to get rid of the duplication to be able to make
this new functionality visible on the beta cluster.
Actual removal will hapen the moment we actually remove the
old toolbar. There are already tickets for this.
Bug: T286765
Change-Id: I842c3c39a55a273af20643fa8a602d2e57fb6b8c
This affects only the new sidebar. Deprecated parameters don't
get a checkbox, except they are used already. "Used" includes
parameters that are present, but empty.
Bug: T274551
Change-Id: I6ebd020d02650c19060345d13495373acab363df
This method already exists in the ve.dm.MWTransclusionPartModel
base class where it does the exact same.
Bug: T274551
Change-Id: I19d5914ed9b4b435c83ea4d64019bc46ce1ce8fd
This reverts commit 0d4dee341b.
Reason for revert: This made it entirely impossible to add a
deprecated parameter, even if done intentionally. Needs more tests.
Bug: T274551
Change-Id: I7389bad0845cd1ce78f9d7ef71592cb1ce2a063e
They both want to handle the same URLs. If we detect that
DiscussionTools new topic tool will open, then do not open 2017
wikitext editor.
Bug: T282204
Change-Id: Ic7dd677ea219938969f60bab91387c2e03ebdbe6
Disambiguation pages are rarely the page users intend to link to,
especially with newbies. By moving the disambig page(s) as the last
result, the user is more likely to pick the page they actually intended
to link to.
Bug: T285510
Depends-On: I2b8545f6dd4849629037f81f48a540748e60da83
Change-Id: Id55a19e7665d8f88559c471de36e5447fb2babb0
If editor loading was aborted while the surface was already being set up,
this code would cause the following exception:
Uncaught TypeError: Cannot read property 'tools' of null
Introduced in c02c529537 (2017) because
our coding conventions at the time demanded that all `var` statements
must appear at the beginning of a function.
Bug: T287487
Change-Id: Id657d6f1e1189c17ede25362f145bb7b10f441db
With this users can also filter undocumented parameters
that they added to an empty starting template.
Bug: T272481
Change-Id: I99adb38b0ae4d4ade91fcb506f10c0222b9bb5e8
Note that this patch alone probably does not make that much
sense. The code executed is pretty much the same. The only
difference is that the empty (!) …ContainerWidget is kept
and re-filled with what might be a completely different
template.
This is not much of a difference to before when the
container was recreated.
This change will make more sense when the container has to
manage more state, e.g. focus states. This state will
survive then.
Change-Id: Ic336d10a595e3e222741a3dc57c1d54639166b7a
This ellipsis was there before we started working on this code,
but was never working properly.
We understand that the CSS was intentionally done like this (as
the comment explains). However:
* We changed the width of the dialog. The old value doesn't
match any more.
* The width is different when the sidebar is expanded vs. when
it is collapsed. Even if we update the number, it won't
always work.
* The 100% work fine in current browsers. I can only assume
this was different back in 2014 when this CSS was written
(see Ia8259e9).
Bug: T285044
Change-Id: I3de2b0ed0b6a05d2b9fa0b325a2b12277564b271
Notably:
* Don't require the model in the new sidebar via dependency
injection, but connect the event handlers later. This is
relevant because we currently create the new sidebar in the
wrong spot. Removing the hard dependency allows us to split
the code and utilize initialize() and getSetupProcess()
correctly. This will be done in a following patch.
* The change event now includes the new position. This makes
it very easy to add this missing feature to the new sidebar.
Also:
* Stop triggering change events when nothing changed. These
events are expensive. They bubble all the way up to the
TransclusionModel, and to all linked
onTransclusionModelChange() handlers.
* Update event documentation to make this more visible.
Bug: T274544
Change-Id: Iafe29f18a6fed14d9c3124c9756aa840886afbbc
Notably:
* Include parameter aliases, labels and descriptions in the
search.
* Don't use a possibly outdated search index, but live data.
* Clear filter when a new checkbox is added.
Bug: T272481
Change-Id: Ie90a803af6178a8bb6de370a0f8e079800d9f8a2
In detail:
* Allow clicks on all elements in the new sidebar. This should
focus the corresponding element on the right.
* Make all elements in the new sidebar tabbable.
* Fix MWTransclusionOutlineTemplateWidget.createCheckbox() to
not need a temporary param object any more.
* Rewrite more code in MWTransclusionOutlineTemplateWidget to
be shorter and easier to read.
* Fix MWTemplateModel.addParameter() to not do way to much
stuff when a parameter already exists.
* Update code documentation.
* Use more specific, less ambiguous variable and method names.
Bug: T274544
Change-Id: Iaf6f7d1b0f7bf0e9b03eb86d01f3eceadece6fe4
Clicks on the left side now focus elements on the right
side.
This patch also simplifies the …ContainerWidget constructor.
The config parameter should only be used for "OOUI things"
that are needed by subclasses and mixins. But the parameters
we have here are not "UI things".
Passing them as config passes them to classes where we don't
know what they do with it. What probably happens is that
some class keeps a reference to the entire config object,
which doesn't have a benefit and possibly blocks garbage
collection.
Bug: T274544
Change-Id: I0c0e4a1ba59dcb43141338ffe939c9c6783e000d
Actually reusing this OOUI mixin gives us a lot of well
developed functionality we need anyway. Most notably proper
event management, e.g. click events.
The number of CSS properties we need to override is managable,
I would argue. Let's see:
* Our buttons are not inline-elements, but should use the full
width.
* No focus-border left and right for the same reason.
* We want much more inner padding.
* We want a stronger hover effect.
* We need to fine-tune the position of the icon. This is
because of the inner padding.
* Need to get rid of a negative margin that's only relevant
for inline-buttons.
I currently feel like the benefits are worth living with
slightly more brittle code. Note that we can undo this change
any time because all this is well encapsulated in this new
class.
Bug: T274544
Change-Id: I33f275a958964d49e803e56bf74a6fa961093da1
This introduces another generic "button-like" class that can
be reused in multiple places in the new sidebar. The main
change in this patch is the "add more information" button
which is now an instance of this new class as well.
This patch also simplifies over-complicated setup code in
related widgets.
Bug: T274544
Change-Id: I0cfe7675d02fdd5c5dc8d9198bb3f4aec9abf397
Before, the new sidebar was hacked in a place where it confused
the BookletLayout logic. This became visible when using the
up/down buttons to move elements in the sidebar.
This new container wraps the new and the old sidebar. It also
uses a temporary color to make it easier to see where one ends
and the other starts.
Bug: T274544
Change-Id: I4e5b40b1d1556886fc85cff9e926a02e4888f032
The two new widgets are pretty trivial now, thanks to the base
class.
Note there is still no code to delete the widgets. That's also
why you will always see a placeholder widget at the top. This
will be fixed with the next patches.
This patch also renames most of the "…TemplateOutline…" classes
to "…TransclusionOutline…" The reason is that these widgets are
not for a single template, but part of the container widget for
a more complex transclusion (i.e. a sequence of multiple
templates and wikitext snippets).
Bug: T274544
Change-Id: If4219b0b8ad4d1969ab1ec5ec4db0728811bab35
The icon and the name of the template are now created by the
base class. This is meant to be reused for other elements
that are not templates.
Bug: T274544
Change-Id: I76bbc0e8c0420e9c6357d093d5f5e1651a0c2719
Reasoning:
* format=json must be the default. Nothing else makes sense in
the context of this code. This should not be a surprise.
* formatversion=2 is only a default when the custom
getContentApi() is used, but not when mw.Api is used. One
might argue that it's safer to always specify formatversion=2.
However, this is not done in other places in this codebase.
It should never be done or always.
* I find it confusing when the action=… is missing. Let's not
rely on this default.
Change-Id: I6ca29f76bffc0849103c5bcff4aaf28fcaaa4c52
This code doesn't do anything but adding an empty <div> to
something that is already a <div>. It doesn't even have a
class name, i.e. it's not referenced from anywhere. We can
add such containers back any time when it turns out we
actually need them.
Bug: T274544
Change-Id: I62546cc7939364db31f37b9de0c035974554544b
This base class will be used to style the 3 types of top-level
items in the sidebar the same way, without the need to
duplicate code or styles.
Bug: T274544
Change-Id: I1a62ff610728d7150dea1717316ef20f6882783a
This matches the existing naming scheme better. I also plan to
re-use this class for other types that are not templates.
That's why the name is the more generic "transclusion" now.
This patch also removes a `padding: 2px` that's not that
helpful. We will need paddings later, but need to choose them
much more carfully.
Bug: T274544
Change-Id: I6f0f630da2230b023b3fb065e5ad86d8211bb7b3
Cloning the #catlinks node loses all native event handlers registered
on it. jQuery's `.clone( true )` method can only copy jQuery event
handlers, so it does not help. As a result, when the node is
reattached to the page after cancelling visual editing, HotCat's
interface is still visible, but not functional.
To fix this problem, keep a reference to the original node rather than
a clone.
https://developer.mozilla.org/en-US/docs/Web/API/Node/cloneNode#noteshttps://api.jquery.com/clone/
Change-Id: I2c0b3d1a919b67053a17dd11fd2b7dc7556267ef
Because the API uses a generator, the search results are ordered
alphabetically. The actual search result ranking is in a .index
field. This code accidentially deleted the alphabetically lowest
template instead of the least relevant one.
Change-Id: I79de024feb569e9f06bedab908a6509a4d4fa99b
We do this additional prefixsearch anyway. What we did before
was ignoring the result when it was not a 100% exact match.
Instead we can always add this 1 prefixsearch result when it
was not already part of the CirrusSearch result set.
This won't happen often. Usually the 1st prefixsearch result
was already part of the CirrusSearch result set anyway. But
if it wasn't, that's a serious issue for expert users that
expect the search to behave similar to the suggester at the
top of the MediaWiki interface (which is also a prefixsearch).
Change-Id: I959d2b058a3d64596a8cfbe5476ab351e40f8760
Some details:
* The config is not optional in these cases.
* This patch continues to remove some comments that don't add
any information but just repeat what the code already says.
Change-Id: I5c27cd01ad80709bb583256821d65c6b65b74b05
Separation of concerns:
* The template model knows which parameters are currently used,
but doesn't know what's documented.
* The spec knows what's documented, but doesn't know what's
currently used.
Change-Id: I97cac00d6775a17a07059d0e8a7a116adc6080b3
There are some methods that behave different when a parameter
placeholder (where the parameter name is "") is present. Some
skip placeholders, some don't. This is critical to cover
before we make further changes to this class.
What I also do in this patch:
* Use shorter variable names to make the code easier to read.
* Don't reuse the `transclusionData` variable but use a copy
of the expected value. This makes the assertions much
easier to understand.
* Bring every test in the same "setup" → "execute" → "assert"
order.
Change-Id: I41a691c56bc509b132dc719ff820ae1ade4ccc3a
For example, checking if a parameter is required works just fine
for unknown parameters. They are never required. Since I16708b0
we don't need to guard the spec related methods any more.
Change-Id: Id90e4cb810dc9faca3b26f122a534f276ee31709
It's good practice to make transparent elements transparent
for mouse clicks as well, i.e. make it possible to select text
behind the fade effect.
Bug: T283943
Bug: T286235
Change-Id: Ib5022a74c70e4b7cb5e2a0faad20bd9abcc0da36
Introduced in 2 separate patches by the same author. This
patch removes the line that was introduced last.
Change-Id: I77575f7afe0f9276c7b54ee44d828e7ccb87c978
Before the method fetchRequestAlways() was doing two entirely
different things. Note how the two function arguments are
split now. Each method uses only one of them.
Change-Id: I592a1f29fd9c677a0ff18115cccda36950172001
These methods are special in so far that they create *minimal*
wikitext where optional whitespace is not preserved. I tried
to rename the methods to reflect this, but could not find a
caller. What's used instead are the .serialize() methods.
Bug: T284895
Change-Id: Iedaa5b7efa9675151cc0553854d8aef3f9a46cbb
A lot of the checks are redundant. The first check still is
redundant because the later two cover everything as well. But
I left it for performance reasons.
Additionally:
* There was no test for the method.
* This patch also updates a few pieces of documentation in the
same class.
Change-Id: I10f2944a844cc070bdc08dec6719929b383e34fa
If a known parameter is present using one of it's aliases, then
only the aliased name should be shown to the user. This patch,
therefore, resolves the issue of the same parameter being added
to the sidebar twice.
When adding a parameter that is aliased, it will receive the same
position as the non-aliased parameter it is replacing.
Bug: T274545
Change-Id: If4e58c941fd0f0e690d3603935f5a5d3f9938163
This also removes a few lines of text that don't explain
anything that would not be obvious from the code or @return
tag anyway.
Change-Id: I2f8f02dd61c50d9990d72c0e8ea79d679c9b11f2
It's allowed in values, but not in parameter names. The moment
a parameter name contains an `=` the parameter name will be cut
off at this point, and what's behind the `=` will become part
of the value.
You can test this on any live wiki. Open VisualEditor. Edit any
template. Add a parameter with a name like `a=` and some value.
Switch to wikitext mode and back. Edit the template. The `=` is
now part of the value.
Bug: T98065
Change-Id: I5e00e8fac987471243605816b041d3638927ac3b
This fixes a minor issue in the spec class. In the first step,
parameters from the template are added to the list of known
parameters. Later, aliases are resolved. The original behavior
was that such a parameter moved to the end of the list. This
is rather unexpected.
This dosn't have much of an impact. The pretty much only place
where the parameter order from the spec can be seen is in the
parameter search widget. Still I believe it's worth fixing.
Bug: T285483
Change-Id: I455818451811e92bba3e9320c2d41e1db8d563f2
I don't want this code to crash when the TemplateData API
returns an unexpected result.
Bug: T285483
Change-Id: I237cbfbb85892a53a08d9e7e34cf4974775d627a
This is just not necessary. It removes a level of indirection
that possibly makes it harder to understand the code. It makes
it easier to possibly get rid of unused methods.
Change-Id: Iaf8b213a5e1ae64a24b5bcdf2a0b200d5d3cbf46
This doesn't "extend". It was never used like this. What it
actually does is to link between a (cached) TemplateData blob
and the spec class that want's to use it.
Is this the best possible name?
* fillFromTemplateData( … )?
* propagateTemplateDocumentation( … )?
* readDocumentationFrom( templateData )?
* …?
Do we want to rename the "spec" class as well?
* MWTemplateDocumentation?
* MWTemplateMetadata?
* MWTemplateDataAccessor?
* …?
Bug: T285483
Change-Id: I6c52ef42d411c2f47fc0080768d36ebda4dd2a55
Just store the JSON blob from the TemplateData API as is.
This comes with a bunch of nice consequences:
* Less code.
* Less class properties that don't do anything but copy what's
in the TemplateData blob.
* Easier to understand what's going on. The `this.templateData`
property is now a reference to the *actual* TemplateData
documentation.
* No need to cache the documentedParamOrder. Just do it when
needed.
This also removes an unused feature from the `extend()` method
that didn't made sense anyway. Before it was possible to merge
conflicting documentations. But this is not only unused, it's
impossible to have multiple documentations for the same
template.
The method acts as a straight setter now. The next patch will
rename it accordingly.
Bug: T285483
Change-Id: I3ffc202577e9a20fc7491234601ccd981113f866
Instead of faking entries in this.params, let's use a separate
tiny data structure to keep track of parameters we have seen so
far, and in which order.
This finally allows to easily distinguish between documented and
undocumented parameters.
Bug: T285483
Change-Id: Idf62b0661178a3bbef7e817edf016dbd572d415b
When what you type is a partial match, you can't add it as an
unknown parameter, even if that would be the correct action. The
reason for this unexpected edge-case is a mistake in the code
where a variable called "exactMatch" is set when a *partial*
"nameMatch" was found.
Bug: T285940
Change-Id: I6d12e2d7251a19d7d5f8be544c3c32a3ac14fcf0
The so called "spec" class keeps track of parameters that have
been used before, no matter if documented via TemplateData or
not. Removed parameters are still "known" (i.e. have been seen
before).
This feature allows to easily find previously used parameters
names when an undocumented parameter was removed and the user
tries to add it again.
Bug: T285483
Change-Id: Ia1555eea87cd99e7a3f386f4279ec5a80fb98a79
I rearranged this piece of code like a dozen times before I
finally understood what it actually does. This should be much
more obvious now.
The idea is:
* If no edit was made the button is always disabled.
* You can save pretty much everything, except when the
transclusion still starts with a placeholder.
* You can also click the done button when the dialog is empty.
This feels a bit odd, but was like this before. I think this
codepath is unreachable. But it probably doesn't hurt to
keep it.
Bug: T284895
Change-Id: Ic483201b64fd64f414c5b1ec4c44198b8eadb9f2
These tags don't do much, if anything. But they provide a hint
in which scope a method might be used.
Bug: T284895
Change-Id: I0b4bdd416ee89d26961c4ded4d8bbace8c57da76
In I04b8a14fbec7be5a1c4defabf92e94f694c1e638 we sepearted params from
aliases. There we missed that re-filling the parameters from the
template could re-add the aliases.
Bug: T285483
Bug: T285843
Change-Id: I1928b443a5f708bc8c57efa5ad0a86b5915b159c
While the term "canonical" is not wrong, I find it still
somewhat ambiguous.
1. "Canonical" could mean different things. E.g. is the order
of parameters as they appear in the article's wikitext the
"canonical" one? It's possible to argue like this, esp. if a
template doesn't have TemplateData documentation. In this case
the only order known is the one from the wikitext.
2. "Canonical" sounds like the parameters must be reordered.
But this should never happen. Not having dirty diffs is more
important than having the parameters in a specific order.
Bug: T285483
Change-Id: I23658d37fea50b727667677ac6a49066673b2135
