The page-issues reducer function that retrieves the severityLevels
for issues was incorrectly comparing two values in the multiple-issues
scenario.
Instead of comparing the severity of the current issue with the previous
value in the accumulator, it was comparing the current issue with the
previous issue in the original array, thus ignoring the
previous `maxSeverity` value in the accumulator.
Tests added and function refactored with more explicit variable names.
Bug: T203725
Change-Id: Ia4c15cbf0c2457d68a7381e8ed04c1a6b3ae65d1
Erroneously in I6fd55c35b9e2ce35894259f36d1a50fb5dca5e43 for the old treatment
we sent sectionNumbers for all issues in the page.
This is inconsistent with the issuesSeverity field above it
Add a clarifying inline comment.
Bug: T203050
Change-Id: Ib1fcda0c49a162cd7aca8ee8b3221236f724e1d7
Rather than using err and error as variable, use error for consistency
Follow up to I07f01b4c025b2e5e4cbf88ec05e7c536442c62cc
Bug: T202026
Change-Id: I54165ff1f1b17284d8232c491244e1a98950d5e2
createBanner was incorrectly always assigning the 'all' keyword
to the old page issues banner. Instead it should use the section
number in the function signature - as this decision of which issues
to show is made inside initPageIssues
For the old treatment, in the main namespace, we only show issues
in the lead section. When we use the 'all' keyword the visual is
the same, but it breaks the instrumentation requirement that
sectionNumbers and issuesSeverity should be the correct length
Note, for the talk and category pages treatment, we do not
log any events so the instrumentation doesn't matter here
and the 'all' keyword correctly targets all issues in the page
as before.
Bug: T203050
Change-Id: I63e45da05ca033fe282633f7fd59038a8e5d8c8d
For the page-issues modalClose event, the number of values for `sectionNumbers`
and `issuesSeverity` should be the same, since `sectionNumbers` should describe
the the section of each visible issue in the modal, not the section of the
modal itself.
Bug: T203050
Change-Id: Ic58c5940a6059e71aa3aeed26232afbe8faf1618
The plugin checks and flags potential security issues (XSS, SQLi, etc.)
using static analysis.
See <https://www.mediawiki.org/wiki/Phan-taint-check-plugin> for more
details.
Change-Id: Ief36c5b7c3fc61950e52044fde7feeed9fe28831
Extensions may be using these tags and not want
these styles (especially the border).
Bug: T203474
Change-Id: I03a22cf6377002f968cabdcce9354e73354fb6b8
When handling special cases that are logically distinct from
the function's main branch, it improves code quality (through
readability and maintainability) to place those first and with
an early return.
The has the benefit of the main return statement being easy to
find at the end of the function. (Not early and/or in a block).
It also means when working on the code, there is generally a
less complexity and fewer nesting levels, given that most code
is in the main branch. This makes is easier and quicker to verify
that code does what it should, as well as making it easy to
extend in the future. When considering to add code to end of a
function's main scope, it should relate to the function's main
branch by default, not a special case. For example, a getName()
method should not end with a top-level statement 'return false'
(unless it is a stub). Rather, one would expect it to end with
`return name`.
Change-Id: I1f3088f2409c82dd3bf757fc8fa27dc97ae2767b
Not calling an explicit output format defaults to ->escaped(), which often
leads to double escaping.
Spotted by the phan-taint-check-plugin.
Change-Id: Ie527768bea670808e63cfc8cbff64015ae29d4a3
This adds a client side error logger that will store errors
in EventLogging. We will use it to get a sense of the number
of errors inside mobile.
To enable make use of the new configuration variable
wgMinervaErrorLogSamplingRate = 1
Notes:
* the optional `meta` field will not be utilised by the generic
error handler.
* URI length is not trimmed for title. We will consider whether we need
to make any adjustments to the schema values during roll out.
** For stack trace, we limit the length by removing errorUrl (which is logged
separately)
Testing:
Add throw new Error('asasasa'); anywhere in your code and make
sure the code executes. It should trigger an Error event provided
that configuration has been updated.
Bug: T202026
Change-Id: I07f01b4c025b2e5e4cbf88ec05e7c536442c62cc
"multiple issues" templates as one issue.
When logging the `issuesSeverity` and `sectionNumbers` field,
any issues that are part of a "multiple issues" template only send
one value.
Adds an `isMultiple` property to IssueSummary to determine which
issues are part of a multiple-issues template.
Bug: T203050
Change-Id: I7d55dfead72439df4accadcdc8623a080e1321c2
The value of `sectionNumbers` should be the section number of each issue
Not the number of the sections that have issues.
Bug: T203050
Change-Id: I6fd55c35b9e2ce35894259f36d1a50fb5dca5e43
MobileFrontend no longer uses z-index:5 for the toobar,
the default z-index:2 is sufficient for the overlays to
appear on top.
Lowering this value allows toolbars to appears on top of
local overlays, such as deactivated selections.
Bug: T202990
Change-Id: I55ff6971249427c12c090018fcd4d4e5d0ec85d1
Adds logging for the sectionNumbers field in the PageIssues schema.
Additional changes:
* createBanner now requires section number to be a string - this ensures
consistency with how these are used.
* fix a bug which meant createBanner was being called with undefined
section number (due to table of contents)
* Fix some indents in some JSDoc blocks
* Change parameter in function signature from mixed type (int or string)
to explicit string
* update schema number
Depends-On: Ia2696b86c6855d7b46a3f668585377d106d7af23
Bug: T202098
Change-Id: I20511a77258ea245f3d6fe93ade238e5df397a71
An event only needs to send this field when overriding the
default. This line is thus unnecessary.
Change-Id: Ia1dfcac5dd25f36f5f4169daf6535fd66aedf754
Previously, the main edit icon in the page actions was handled indirectly
by changing the URL and then routing the edit via detecting the hash
fragment has changed. On the other hand section edit links were hijacked
via a click handler. The latter is better as it doesn't modify the DOM,
which allows the editor to override the JavaScript behaviour. It's also
preferable as it doesn't interfere with EventLogging - this subtle difference
was supressing edits to the Edit and PageIssues schema.
Bug: T202786
Change-Id: I4175bc6f0ddda28397d185502d1839716d051c56
