Now, there's just one spot at the beginning of the function where we
inspect the string key the user gave us; and after that point, we not
only have validated that string but in fact are working from our own
record that it pointed to, not the string itself.
This simplifies the code a bit, e.g. by not repeatedly searching the
database for the key (and hoping everything agrees so that we keep
getting the same row), and it will simplify adding logic to inspect
row attributes like `presume_email_valid`.
There's no use case for presenting a key that's invalid; if we haven't
given the user a valid key, we needn't send them to a URL that
presents an invalid one. And the code is simpler to think about if
the only keys that can exist (after the validation at the top of the
function) are valid ones.
Apart from the case where creation_key is None, but invalid, and
settings.OPEN_REALM_CREATION is True so that we'd previously let the
invalid key slide, this is a pure refactor.
This fixes an issue where the user's own avatar was being sent down
the wire as None. We could have fixed it, as in #8265, by adding code
in the webapp and mobile apps to compute medium-size gravatar URLs as
well, but that would be messy, and there's little benefit to that
complexity (saving at most 2 URLs from the payload).
Fixes#8253.
Different formats for configuration files have a wide variety of ways
of representing lists; so if you're not accustomed to Python syntax,
or aren't thinking of this file as Python code, the syntax for several
ALLOWED_HOSTS entries may not be obvious. And this setting is one
that an admin is likely to want to touch quite early in using Zulip.
So, demonstrate a multi-element list.
For similar reasons, demonstrate an IP address. This one is in a
range reserved for documentation (by RFC 5737), like `example.com`.
Fixes improper katex rendering in markdown help template. The
rendering was originally correct, but was broken when we upgraded to a
new version of KaTeX with a different HTML/CSS interface.
"Sign up" -- pretty important for people to be able to read that!
There's also the "or" that gets rendered as a big "OR" in a divider --
but that styling is too brittle, so that if "or" becomes e.g. "oder",
the lines stay the length they should be for "or" and the lot of them
wrap. Yuck. Better to leave it as "or" until that's fixed.
In password modal, submit button's data-dismiss property should be
removed, as on data-dismiss we flush all data of form fields,
which results in passing empty string to json request.
We already flush all data of fields after compeletion of request.
This restores the property that changing one's name in on browser's
"account settings" also changes the user's name in other browser windows'
"account settings" pages.
This makes this UI widget more consistent with its neighbors.
tabbott: This introduces a bug with how the `full_name_field` HTML is
managed; it should be done via the `server_events.js` handler.
- When password fields are cleared, update password strength bar.
- On data dismiss, clear password fields.
- Exclude forgot-password div from focus.
- On enter key, modal should be submitted not dismissed.
- Change password btn text from "Change password" to "*****"
This makes many diffs that touch our JS dependencies in package.json
much, much more readable by suppressing multi-thousand-line changes.
With that kind of change routine, it's clear this isn't a file people
are closely reviewing the text diffs in anyway.
This script and our CI scripts tools/travis/{backend,frontend} have
stayed pretty well in sync in the 6 months since 360c27ded made that
relationship explicit and easy to check!
Just one small exception; so fix that.
This may or may not be temporary, but either way, the other code is
there in source control, and the "why" of disabling gitlint is the
helpful bit for a comment.
Greg Price
Tim Abbott
rht
AmAnAgr
Shubham Dhama
Robert Hönig
Tarun Kumar
YJDave
Balaji2198
Vishnu Ks
Umair Khan
Brock Whittaker
Mukul Agrawal