`ply`, [by design](https://github.com/dabeaz/ply/issues/79), does not produce reproducible table files; hence bug 1633156. (Note that this was *always* true, but only became a problem once we switched to Python 3, which has more unpredictable dict iteration order than Python 2.7, at least prior to [3.7](https://docs.python.org/3/whatsnew/3.7.html#summary-release-highlights).)
In any other circumstance I would consider submitting a patch to `ply` to fix this, but as of the [in-progress version 4.0 of the library](https://github.com/dabeaz/ply/blob/master/CHANGES), it doesn't even emit this cached data any more, and indeed the [latest version of the code](1fac9fed64/ply) doesn't even call `open()` at all except to do logging or to read the text data to be parsed from `stdin`. So if we were going to pin our future on `ply` and upgrade to later versions of the library in the future, we would have to live in a world where `ply` doesn't generate cached table files for us anyway.
Emitting the cached table files so later build steps can consume them is an "optimization", but it's not clear exactly how much actual value that optimization provides overall. Quoth the `CHANGES` file from that repository:
```
PLY no longer writes cached table files. Honestly, the use of
the cached files made more sense when I was developing PLY on
my 200Mhz PC in 2001. It's not as much as an issue now. For small
to medium sized grammars, PLY should be almost instantaneous.
```
In practice, I have found this to be true; namely, `./mach build pre-export export` takes just about as long on my machine after this patch as it did before, and in a try push I performed, there's no noticeable performance regression from applying this patch. In local testing I also found that generating the LALR tables in calls to `yacc()` takes about 0.01s on my machine generally, and we generate these tables a couple dozen times total over the course of the `export` tier now. This isn't *nothing*, but in my opinion it's also not nearly long enough where it would be a concern given how long `export` already takes.
That `CHANGES` file also stresses that if caching this data is important, we have the option of doing so via `pickle`. If and when we decide that re-enabling this optimization is valuable for us, we should take control of this process and perform the generation in such a way that we can guarantee reproducibility.
Differential Revision: https://phabricator.services.mozilla.com/D73484
Instead of manually defining toStringTag we now add the toStringTag symbol to the list of properties.
This is also how we usually define toStringTag in the JS engine.
Even though this changes more code I like this approach better. Everything is centralized in the generated bindings file.
Differential Revision: https://phabricator.services.mozilla.com/D72179
Instead of something like
"ByteStringSequenceSequenceOrByteStringByteStringRecord" we should have
"(sequence<sequence<ByteString>> or record<ByteString, Bytestring>)" in error
messages.
Please take a careful look at the resulting error messages (see attachment
in the bug) to see whether this makes sense.
Differential Revision: https://phabricator.services.mozilla.com/D64890
--HG--
extra : moz-landing-system : lando
Instead of something like
"ByteStringSequenceSequenceOrByteStringByteStringRecord" we should have
"(sequence<sequence<ByteString>> or record<ByteString, Bytestring>)" in error
messages.
Please take a careful look at the resulting error messages (see attachment
in the bug) to see whether this makes sense.
Differential Revision: https://phabricator.services.mozilla.com/D64890
--HG--
extra : moz-landing-system : lando
Dictionaries that we never initialize with JS values don't need a full-featured
Init() method. Instead, we output a cut-down Init() method that doesn't even
take a JSContext and Value as argument, and skips as much work as it can. It
uses constant-false for "is the value present?", but also, to avoid compilation
errors due to use of `cx` and `val` in now-dead conversion code, it tells the
native-to-JS conversion machinery that the value is always missing, which lets
it skip most of the the work it would normally try to do and just output
initialization to the default value. We only need to do this for members that
have default values; the others either remain no-passed or are required members
with no default-initialization behavior.
This saves about 330KB of codesize on Linux64 without PGO and 285KB with PGO.
Differential Revision: https://phabricator.services.mozilla.com/D48007
--HG--
extra : moz-landing-system : lando
This saves about 270KB of codesize on Linux64 without LTO, or 20KB with LTO.
The basic idea is that we can flag dictionaries that need to-JS conversion
(hence ToObjectInternal) based on various IDL uses (return value in normal
interface, argument in callback, etc) and then annotate the ones that are
converted to JS manually in C++ code.
The mozwebidlcodegen changes are needed because non-local changes (e.g. whether
a dictionary is used as a return value somewhere) can now affect the code
generation for a dictionary and hence whether the relevant binding file should
be regenerated. Since these changes can happen in any .webidl file, we need to
check for them. We can't track this via the dependency set on the dictionary
itself, because that would not notice new uses being added.
Differential Revision: https://phabricator.services.mozilla.com/D48006
--HG--
extra : moz-landing-system : lando
This saves about 200KB of codesize on Linux64 without LTO. No effect with LTO,
but is needed for the following patches to work.
Very few dictionaries need these conversions, so explicit opt-in is fine.
Differential Revision: https://phabricator.services.mozilla.com/D48005
--HG--
extra : moz-landing-system : lando
The changes to the IDL files were done by running this in dom/webidl:
perl -pi -e 'BEGIN { $/ = undef; } s/\[HTMLConstructor,\n Exposed=Window\]\ninterface ([A-Za-z]+) : HTMLElement \{/[Exposed=Window]\ninterface \1 : HTMLElement {\n [HTMLConstructor] constructor();\n/g' *.webidl
and then fixing any remaining parser failures. That involved hand-editing the
following files:
TestCodeGen.webidl
XULFrameElement.webidl
XULMenuElement.webidl
XULTextElement.webidl
XULTreeElement.webidl
HTMLAudioElement.webidl
HTMLDialogElement.webidl
HTMLElement.webidl
HTMLEmbedElement.webidl
HTMLFormElement.webidl
HTMLImageElement.webidl
HTMLObjectElement.webidl
HTMLOptionElement.webidl
HTMLSlotElement.webidl
HTMLVideoElement.webidl
XULElement.webidl
XULPopupElement.webidl
Differential Revision: https://phabricator.services.mozilla.com/D49349
--HG--
extra : moz-landing-system : lando
For review purposes, the important changes are in dom/bindings/Configuration.py
and dom/bindings/parser.
The changes to the IDL files were done by running these in dom/webidl
and dom/bindings/test:
perl -pi -e 's/^interface ([A-Za-z0-9_]+)($| [:{])/[Exposed=Window]\ninterface \1\2/' *.webidl
perl -pi -e 'BEGIN { $/ = undef; } s/\[HTMLConstructor\]\n\[Exposed=Window\]/[HTMLConstructor,\n Exposed=Window]/g' *.webidl
perl -pi -e 'BEGIN { $/ = undef; } s/\[NoInterfaceObject\]\n\[Exposed=Window\]/[NoInterfaceObject,\n Exposed=Window]/g' *.webidl
perl -pi -e 'BEGIN { $/ = undef; } s/\[ChromeOnly\]\n\[Exposed=Window\]/[ChromeOnly,\n Exposed=Window]/g' *.webidl
And running this in dom/chrome-webidl:
perl -pi -e 'BEGIN { $/ = undef; } s/\[ChromeOnly\]\ninterface/[ChromeOnly, Exposed=Window]\ninterface/g' *.webidl
and then fixing all the resulting parser failures. I then verified that the
generated code is the same as before this change.
Differential Revision: https://phabricator.services.mozilla.com/D46697
--HG--
extra : moz-landing-system : lando
There are no callers who want this. They all filter out external interfaces in various ways; we might as well do it up front.
Differential Revision: https://phabricator.services.mozilla.com/D46695
--HG--
extra : moz-landing-system : lando
A bunch of loop-detection, etc, complexity goes away because mixins are not
interfaces and the mixin syntax does not allow various things we had to guard
against in terms of maplikes and whatnot.
Differential Revision: https://phabricator.services.mozilla.com/D46524
--HG--
extra : moz-landing-system : lando
We can remove this once [PrimaryGlobal] goes away, but for now we need to not
have this issue, which actually bit me when converting NavigatorLanguage to a
mixin.
Differential Revision: https://phabricator.services.mozilla.com/D46510
--HG--
extra : moz-landing-system : lando
The grammar changes parallel those in https://github.com/heycam/webidl/pull/700
We don't prevent having both a constructor operation and [Constructor] or
[ChromeConstructor], because those extended attributes are about to get
removed, once they are no longer used in our IDL.
Differential Revision: https://phabricator.services.mozilla.com/D45387
--HG--
extra : moz-landing-system : lando
Add IDLInterfaceMixin with a new superclass shared with existing IDLInterfaceOrNamespace.
Differential Revision: https://phabricator.services.mozilla.com/D38802
--HG--
extra : moz-landing-system : lando
Ricky Stewart
Tom Schuster
Nika Layzell
Bogdan Tara
Anthony Ramine
Boris Zbarsky
Arthur Iakab
Edgar Chen
Emilio Cobos Álvarez
nordzilla
Oana Pop Rus
Kagami Sascha Rosylight
Coroiu Cristina
Henri Sivonen
Dzmitry Malyshau