* Minify stream, add stream stuff to releases again
* Kill off a lot of tech debt, drop internal utilities from npm
1. Kill `module/`, internalize `bundler/`, privatize `test-utils/`
We've been telling people to move elsewhere from these for a while, and
it's about time we just pull the plug here and finally remove them.
- We officially removed the bundler from the public API in v2.0, and
that was the only one of these that was ever publicly documented.
Usage should be low enough by now it shouldn't break anyone- I'm not
seeing bundler bugs being reported anymore, either.
- The `module/` utility was so narrow and caveat-filled that I'm not
sure anyone really used it (even us core Mithril devs never really
used it), and we only had it documented in the repo folder it lived
in. I think only one bug was ever filed, and it's because it somehow
ended up completely non-functional without any of us realizing it.
- The test utilities were meant to be internal from day 1, but people
started using it despite us core developers constantly telling people
to look elsewhere and even the docs recommending specific alternatives
without mention of our internal mocks. (Now if people would RTFM,
that'd be nice...)
2. Add dedicated HTML test files to verify ospec and the promise
polyfill, and ensure the promise tests are in pure ES5.
These are made specially for those and should be much easier to just run
now.
3. Fix the benchmark script to use the real DOM in browsers and to not
require as many dependencies to create. Also, tweak them to be much
more effective and precise on what's being tested.
Previously, it was rendering to the HTML file itself, while now it's
rendering to the `body`. This means in browsers, it's triggering layout
and everything, benchmarking how well Mithril optimizes for style and
layout recalcs, too. It also puts some pressure on the hyperscript
parser attribute application, so that can be noticed as well.
* Update dependencies
Also, I normalized them to all be sentences for consistency, and I moved
the reentrancy check from `m.mount` to `m.render` to be a little more
helpful. The router change during mounting is inconsequential and only
to avoid the new modified error, and the change to the update loop is to
send the original error if an error occurred while initializing the
default route. (This is all around more useful anyways.)
And while I was at it, I fixed an obscure bug with sync redraws.
* Fix assertion descriptions
Move return statement to the end of define()
* ospec: Fix assertion definitions
* Fix typo in assertion
* Add test for descriptions being returned on fail
* Reference result instead of self in returned description method
* Fix style errors
* Actually return the check from `maybeSetContentEditable`
Lots of code paths relied on it being a boolean. When I created the
abstraction, I apparently forgot to make sure it returned the result.
* Don't forget to copy instance state over
* Update changelog [skip ci]
* Fix changelog issue [skip ci]
- Lot of people couldn't migrate to v1 and plan to reevaluate when v2 is
released.
- It's "npm" not "NPM". It doesn't stand for anything, and it never
has - it was initially chosen simply because it was easy to type.
It has a lot of unofficial backronyms with "Node Package Manager"
being one of the most common ones, but it's never officially stood
for anything as an acronym *or* initialism.
- Fixed a few errors in the change log, like non-breaking changes being
included in the "Breaking Changes" section and an inaccuracy in the
summary of a particular change.
- Fixed RawGit URLs to point to GitHack, which is a lighter proxy that
offloads caching to Cloudflare instead of also implementing it itself.
(It also just uses nginx for all the important server logic, so it
scales better.)
- Add a few more v0.2 references as appropriate
* Fix#2414, address part of #1687
Also cleared the CSS up to be a lot more readable instead of smooshed
into a single line.
* Redo the testing docs page
- Addresses another part of #1687
- Also, fix a few linter issues in the ospec binary
* Add note about third-party cookies, tweak a line
* Make the JSX comparison much more meaningful
And let the code speak for itself. Don't fuel the flame wars any more
than what they've already become. We should be *unopinionated*, and so
I've updated those docs to remove the existing opinion.
* Remove a bunch of outdated ES6 references
* Remove the CSS page
* Fix a copy/paste fail
Also, fix some incorrect tests.
* Clarify how routes are diffed, improve key + route resolver docs
- Add some missing links to route resolvers and single-child keyed
fragments, clarify usage around them.
- Drive-by: remove a redundant sentence that itself was missing a
period.
* Actually test for propagation and preventDefault
Previously, the mocks were both junk and inaccurate. No wonder my tests
were silently failing - they were wrong and not obviously wrong.
* Recast the router API to be a lot more intuitive.
Fixes#2387Fixes#2072
Fixes quite a few issues reported on Gitter.
For `m.route.Link`:
- More intuitive
- More accessible
- More ergonomic
- It can be disabled
- It can be cancelled
- It can be changed
- Oh, and you can use it isomorphically.
For `m.route.prefix`
- You can *read* it.
- You can write to it, of course.
- It's literally just setting a property.
For the router itself (and the rest of Mithril):
- You can now `require("mithril")` and all its submodules without a DOM
at all. There is a catch: you can't instantiate any routes, you can't
mount anything, and you can't invoke `m.render` in any capacity. You
can only use `m.route.Link`, `m.route.prefix`, hyperscript stuff, and
`mithril/stream`, and you can use `m.request` with `background: true`
if you use a global XHR polyfill. (You can't use `m.request` without
`background: true` except with a DOM to redraw with.) The goal here is
to try to get out of the way for simple testing and to defer the
inevitable `TypeError`s for the relevant DOM methods to runtime.
The factory requires no arguments, and in terms of globals, you can
just figure out based on what errors are thrown what globals to
define. Their values don't matter - they just need to be set to
*something*, even if it's just `null` or `undefined`, before Mithril
executes.
Had to make quite a few other changes throughout the docs and tests to
update them accordingly. Oh, and that massive router overhaul enabled me
to do all this.
Also, slip in a few drive-by fixes to the mocks so they're a little
easier to work with and can accept more URLs. This was required for a
few of the tests.
* Update changelog + numbers, add forgotten bundle option
* Add PR numbers to changelog [skip ci]
* Allow continuing to the next match by returning `false`.
* Update numbers again
* Drop `m.version`
It's caused way too much grief over the years, and I've finally decided
it's worth pitching. For those who need it, it's easy to get, especially
if you use it through Node or a build system. And for those who are just
loading it globally, you have to explicitly specify the version anyways,
so you'd be just as golden if you followed it up with a simple inline
script that does `m.version = "the version you loaded"`.
Oh, and also, you shouldn't be coding specifically for version numbers,
either - it's a known anti-pattern. Instead, you should prefer feature
detection and just do the right thing.
* Update changelog [skip ci]
* De-servicify router (mostly)
Still uses the redraw service, but it no longer has an intermediate
service of its own.
Also, did a *lot* of test deduplication in this. About 30-40% of the
router service tests were already tested on the main router API instance
itself.
Bundle size decreased from 9560 to 9548 bytes min+gzip.
* Merge `m.mount` + `m.redraw`, update router
Simplifies the router and redraw mechanism, and makes it much easier to
keep predictable.
Bundle size down to 9433 bytes min+gzip, docs updated accordingly.
* Make `mithril/render` just return the `m.render` function directly.
* Deservicify `m.render`, revise `m.route`
- You now have to use `mithril/render/render` directly if you want an
implicit redraw function. (This will likely be going away in v3.)
- Revise `m.route` to only `key` components
* Add `redraw` to `m.render`, deservicify requests
* Test error logging
* Update docs + changelog [skip ci]
- Babel 7 has a whole different module API
- Traceur is basically dead
- JSX != HTML
- Some bits are just obvious from context what they are
- ES6 docs shouldn't be auto-installing JSX support
- Also, I decided it was worth clearing up some misleading docs in
surrounding areas.
Nobody has asked yet, but I'd like to not form assumptions about the
reader. Also, it's come up on rare occasion a while back, on both GitHub
and Gitter.
- Remove appropriate route change subcriptions when a root is removed
via `m.mount(root, null)`.
- Don't pollute `onpopstate` and friends - use standard event listeners
instead.
- Simplify and streamline subscriptions, in preparation of adding a
`remove` parameter to `m.mount`.
- Change the redraw internals to redraw immediately, with ability to
cancel via returning a sentinel.
- Change `"bleeding-edge"` for `m.version` in `next` to instead just be
the latest `m.version`. (If you're using `next`, you should know what
you're in for.)
- Update tests to be aware of these changes. (Some were failing for
subtle reasons.)
- Drive-by: remove some uses of `string.charAt(n)` and use `string[n]`
instead.
* Fix#2434
* Treat holes as unkeyed, normalize boolean/null/undefined
This brings a lot better consistency with that API, even though it's
slightly breaking. (I had to update a bunch of tests to correspond with
it.)
* Fill in PR number [skip ci]
* Clarify pathname docs, follow spec with fragments
- Valid URLs must not contain a `#` within its fragment.
https://github.com/MithrilJS/mithril.js/issues/2445
- Our docs were a little confusing and misleading - `m.pathname` isn't
aware of URLs, just path names.
- Removed the relevant extension to `m.parseQueryString` required to
support the hash parsing extension. Now we just shave it off and
ignore it.
- Fix support for arbitrary prefixes, so prefixes like `?#` are
handled correctly.
- Add a bunch of tests to cover various areas of confusion and unusual
edge cases.
* Update with PR [skip ci]
* s/xhr/request/g
`m.xhr` was a relic of the rewrite days prior to the release of v1.0.0,
before it was renamed `m.request` to align with v0.2.x. This just strips
some of that legacy naming.
* Make this work with `async`/`await` correctly.
It looked like a V8 bug, but read the two big code comments and follow
their links. It's a bit more subtle than it looks, and V8's in the right
here.
* Fix for #2423. request.data is replaced by body, params is used for querystring interpolation.
* Updated documentation after code review by shadowhand and isiahmeadows
* Convert indentation to tabs
* Replacing m.request.data and m.jsonp.data with params or body.
* Update request.md
Co-authored-by: Isiah Meadows <contact@isiahmeadows.com>
