Bug 5803 – Inaccuracies in the specification of the grammar

Status
NEW
Severity
normal
Priority
P3
Component
dlang.org
Product
D
Version
D2
Platform
Other
OS
Windows
Creation time
2011-03-30T23:03:45Z
Last change time
2024-12-15T15:21:46Z
Keywords
bootcamp, spec
Assigned to
No Owner
Creator
Rainer Schuetze
See also
https://issues.dlang.org/show_bug.cgi?id=20503, https://issues.dlang.org/show_bug.cgi?id=20504
Moved to GitHub: dlang.org#4014 →

Comments

Comment #0 by r.sagitario — 2011-03-30T23:03:45Z
The language definition has a number of smaller and larger inaccuracies. You can find a comparison chart of the grammar in the docs and a grammar used to parse the public test suite and the runtime library (minus some deprecated features) here: http://www.dsource.org/projects/visuald/wiki/GrammarComparison This is with respect to the state of the documentation in December 2010, so a few additions and modifications have already made it into the current documentation.
Comment #1 by clugdbug — 2011-03-31T03:05:48Z
One fairly big mistake in there: there seems to be confusion between StructLiteral and StructStaticInitializer. Foo z = { x: 7, y: 56}; is a struct static initializer, not a struct literal. Foo(7, 56); is a struct literal.
Comment #2 by hoganmeier — 2011-03-31T04:22:11Z
I'm already working on folding some of the changes I agree with into my github fork @ https://github.com/Trass3r/d-programming-language.org Will push soon.
Comment #3 by r.sagitario — 2011-03-31T12:01:27Z
(In reply to comment #1) > One fairly big mistake in there: there seems to be confusion between > StructLiteral and StructStaticInitializer. Maybe I oversimplified here, but I was hoping for a generalization of the array/struct initializer and literals. IIRC the StructStaticInitializer is said to be deprecated anyway, but it might actually be possible to interpret it as a literal inside expressions with implicite conversions, similar how dmd deals with arrays, because the parser cannot decide which type of array a variable is declared as until the semantic pass.
Comment #4 by clugdbug — 2011-03-31T13:11:27Z
(In reply to comment #3) > (In reply to comment #1) > > One fairly big mistake in there: there seems to be confusion between > > StructLiteral and StructStaticInitializer. > > Maybe I oversimplified here, but I was hoping for a generalization of the > array/struct initializer and literals. > > IIRC the StructStaticInitializer is said to be deprecated anyway, but it might > actually be possible to interpret it as a literal inside expressions with > implicite conversions, similar how dmd deals with arrays, because the parser > cannot decide which type of array a variable is declared as until the semantic > pass. They have very different semantics. Given only a StructLiteral, you can always work out what type it is. But with a static initializer, you know nothing. This might be the only thing in the language like that. It's not implicit conversion, they don't have an initial type to convert from. So auto x = StructLiteral(); works, but auto x = structStaticInitializer cannot. Consequently, you can use a struct literal almost anywhere you can use a struct variable. I've tried to get them deprecated but I don't know if that will happen. They are one of the most complicated things in the language, and the implementation is *full* of bugs.
Comment #5 by bugzilla — 2019-12-17T18:04:33Z
Is this still valid?
Comment #6 by r.sagitario — 2019-12-17T19:44:52Z
I haven't looked through the full diff, but the very first two lines are still valid: both the module declaration and the other declarations are optional which isn't covered by the grammar found in the language spec.
Comment #7 by robert.schadek — 2024-12-15T15:21:46Z
THIS ISSUE HAS BEEN MOVED TO GITHUB https://github.com/dlang/dlang.org/issues/4014 DO NOT COMMENT HERE ANYMORE, NOBODY WILL SEE IT, THIS ISSUE HAS BEEN MOVED TO GITHUB