Merge fork-arnopaehler about qmlweb

[Plaristote]

I could start sorting this out if that's ok. I've already had a look.
There's some coffeescript in there. I'm not sure mixing javascript and coffeescript really is worth it, but if we chose to do it, we'll have to use gulp-coffee.

[ChALkeR]

I would prefer not to pull in coffeescript and keep everything written in js.

es6 would be good, though, and I plan using babel to transplit things, but before that we should merge what we can (increasing code differences would make it harder) and then clean things up.

[Plaristote]

Sounds good.

I started #23, it should take care of the 4617ec4, e20834e, 4e26e05 and e4f6de9 commit.

• 4617ec4 contains examples, and 4e26e05 makes a change to the readme that's irrelevant now. We can reject them.
• e4f6de9 only changes the compiled javascript.
• e20834e is merged in the pull request, with the updated compiled javascript.

[Plaristote]

In 3647358, there's a polyfill for the bind function. If we can all agree with that, I think it would be better not to implement polyfills in qmlweb's code.
Compatibility with older browsers should be the job of libraries such as Modernizr, that qmlweb's users can include to their project if they are concerned with older browser support.

[ChALkeR]

In 3647358, there's a polyfill for the bind function.

That polyfill was actually introduced in 551f92c and removed in f33d663, I agree that we definitely don't need it in the qmlweb itself.

I'm not sure if we should do anything with 3647358 — it basically moves code around and cleans up things. Moving code around atm is undesired (until everything else is merged), minor cleanups were already made, tests are fixed in #25. Are there any actual changes in the src/ folder that are not related to codestyle in that commit?

[ChALkeR]

I started #23, it should take care of the 4617ec4, e20834e, 4e26e05 and e4f6de9 commit.

Great, thanks! Will take a look now.

[Plaristote]

I agree about 3647358, merging it would only make further merge harder. I've taken a look, and I can confirm than apart from codestyle changes, it only adds file that we wouldn't want in this repository anyway, and removes files that have already been removed in master. We shouldn't merge it.

[henrikrudstrom]

checked off some commits i believe are only related to code style and documentation,
uncheck them if you disagree.

[henrikrudstrom]

We dont want the widget and screen class right?

[henrikrudstrom]

We could consider the TextEdit although its very incomplete.
201e20d might be something?

There is a ComboBox implementation, QtQuick.Controls should be a different package i think.
there is also one here: also from @pavelvasev here

[ChALkeR]

@henrikrudstrom

There is no diff for issue texts, so could you also leave a comment text time, please? I see which ones you did check, though, but

There is no diff for issue texts, so could you also leave a comment text time, please? I see which ones you did check, though, and it should be fine this time.

[ChALkeR]

@henrikrudstrom I do not agree with some of these, btw.
The problem is that these are unsorted and blended — some of the commits that mostly reorder code also change classes that were newly added in this branch. Most probably I will have to split these commits into sub-commits.

[ChALkeR]

@henrikrudstrom

We dont want the widget and screen class right?

Actually, why not? But we need to either choose one of implementations from various branches or merge them.

[ChALkeR]

@henrikrudstrom Thanks for the help here, btw!

[henrikrudstrom]

@ChALkeR was late last night: The widget class is actually an example, and yeah, the screen class we should keep.

[henrikrudstrom]

Compared fork-arnophaeler to master,
here are some notable changes, but i dont think any of this be added
QMLBinding.js

Component.js

parser.js

not sure how accurate githubs compare function is, QtWindow/Screen.js was not shown in the diff for example...

[ChALkeR]

Ok. The leftover commits here mostly have invalid commit messages and are mixed up.
Perhaps I would just squash those and split again, like I'm doing with @igosoft patches.

[Plaristote]

I just took a look at this one:

faaaba5 extensive documentation changes code rearrangement in JS — there are some actual code changes here, most probably this has to be split.

I haven't seen any code changes: the biggest visible change I've seen is a modified console.log message. Anything else is just fixing indentation and moving code around. Not sure this is the kind of thing we want to merge, since eslint will help us do this kind of stuff in the future.

There are new comments for documentation (and other removed comments which, I think, could've been useful). It might be useful, but I'm not used to generate documentation from code, so I can't really say.

[ChALkeR]

@Plaristote Thanks, it looks like you are correct and I was mistaken — there are no actual changes in faaaba5, just moving code around.

I don't think that we should merge documentation refactoring now, especially at the current state of master and this documentation. Note how it removes some important comments from function bodies.

We could redo the documentation later, and perhaps we could take a look at these changes when doing so, but I really doubt that someone would wish to take this commit and make it ready for merging on master.

I marked it as rejected.

[Plaristote]

201e20d: most of the changes are made on stuff that were removed from master.
Another change is the walkfixer, which adds code that works around missing walkers (num, array and string). However, these walkers have been added in master. I'm pretty sure walkfixer, and what it becomes in further commits (walk1), shouldn't be merged.
From that commit, only one change seems to be mergeable: the addition of "array" to the constructors object. I've done it in #149.

d4baf53: same here, it changes stuff from 201e20d, imo there's nothing to merge from this commit.

I've also looked at 4086cb0, there's indeed a few things to merge (I see at least tree commits emerging from this, one for adding the Canvas item, one for the gulpfile update, one for the Loader item update, although I'll have to first understand what the last two are actually doing before I merge those).

[ChALkeR]

Ok, updated list of things left here: 9b86ec7, e6aec16, 3647358, a8427a2.

I squashed the top few commits into a8427a2, see the fork-arnopaehler-reordered branch.

[ChALkeR]

Ok, I squashed everything left here into two commits: 8f08b35 and a8427a2.

8f08b35 contains 9b86ec7, e6aec16, 3647358 minus some surely unwanted changes, see the fork-arnopaehler-9b86ec7-reordered branch (its head is even with 9b86ec7).

a8427a2 contains c839ec4, dcda2c3, b83f5ae, 81dba7f, d4baf53, 855c7fd, 201e20d, 4086cb0, 4627a6d minus some surely unwanted changes, see the fork-arnopaehler-reordered branch (its head is even with c839ec4).

[ChALkeR]

The next step would be to split those two and move the style changes out of those. Will do that a bit later.

[ChALkeR]

To check the current status, see the top commits (ignoring [rejected]) at fork-arnopaehler-reordered and fork-arnopaehler-9b86ec7-reordered branches (the ones edited by me).

[ChALkeR]

I will clean this up a bit, removing unwanted changes (files out of src/ and codestyle changes), and then we can discuss what would be left here.

[ChALkeR]

Ok, the actual changes that were not yet applied or rejected are at 92eb17d and 6830c35 now.

I'm going to squash those two now, the next step would be to split those into individual patches.

Update: no, squashing those is not a good idea. Those have to be split separately.

[ChALkeR]

First set of patches (92eb17d split): fork-arnopaehler-patches1.
Commits:

• 5252956 — Implement logCss() function
• a712120 — Implement TextEdit
• 5feb0bb — Implement TestPlugin
• 79c6678 — .gitignore: Add .idea/ and /.project

[ChALkeR]

Second set of patches (6830c35 split): fork-arnopaehler-patches2.
Commits:

• 2585cb5 Row: register effectiveLayoutDirection
• fb6cd39 Add Qt.Align* constants
• e7e6523 Implement Screen
• 384b250 parser: walker updates
• 0a09a8c More work on Loader
• 41f545e A clarification of a TODO comment
• ba8b9ff Register Array as 'array' basic QML type
• 8fe2ad9 Implement TextField
• 79e6124 TextArea: use strict comarison in updateValue
• 1bded10 Implement objItems() and typeName()
• e7cee0d Default to 'Object List' in objList() title argument
• 36f6b2b SequentialAnimation: remove debug, remove descr() debug function
• 501a197 Implement ComboBox
• 3b9fe97 Item: use overflow: auto istead of overflow: hidden
• 954894b QMLEngine: directly assign e.keypad in keyCodeToQt(e)
• 43bccf8 Move Button to QtQuick.Controls
• 55bd5e8 Implement Canvas
• 8cdd241 [ignore] Remove a Canvas TODO stub
• c9def86 Add a test signal to TestPlugin

[ChALkeR]

Closing in favor of #163. Everything not yet merged or rejected from here should be present there.