Bug 16602 – Implicit string concatenation and precedence of ~

Status
NEW
Severity
enhancement
Priority
P4
Component
dlang.org
Product
D
Version
D2
Platform
All
OS
All
Creation time
2016-10-07T19:37:44Z
Last change time
2024-12-15T15:23:56Z
Assigned to
No Owner
Creator
Andrej Mitrovic
Moved to GitHub: dlang.org#3960 →

Comments

Comment #0 by andrej.mitrovich — 2016-10-07T19:37:44Z
In the upcoming version of DMD implicit concatenation of strings is deprecated. However, simply adding ~ may not lead to the same semantics as before. Observe: ----- import std.stdio; import std.string; void main ( ) { string a = "some %s format " "string".format("nice"); writeln(a); // ok string b = "some %s format " ~ "string".format("nice"); // exception thrown writeln(b); } ----- In the second example format() is called before the two strings are concatenated.
Comment #1 by issues.dlang — 2016-10-07T22:52:12Z
Okay. What's the bug? There has been talk of making it so that expressions are always evaluated left-to-right, but AFAIK, it's still the case the order of evalutation is undefined. So, ~ is behaving perfectly normally. You might as well complain that import std.stdio; int foo(int i, int j) { return i * j; } void main ( ) { writeln(42 + 10 * 19); writeln((42 + 10) * 19); writeln(42 + 10.foo(19)); } prints 232 988 232 instead of 232 988 988 With the implicit string concatenation, it basically treated the strings as one giant string, and concatenated them at compile time. ~ on the other hand would only be compile time if it's optimized to be that way. So, they've never had the same semantics, and it doesn't make sense for them to. So, while I grant you that this shows that simply replacing an implicit string concatenation with an explicit one is not necessarily the correct fix for code that uses implicit string concatenation, I don't see how that's a bug.
Comment #2 by andrej.mitrovich — 2016-10-07T23:01:56Z
Right, but I think we should add this note to the changelog. I'm changing the issue to be an enhancement.
Comment #3 by issues.dlang — 2016-10-08T07:24:31Z
I totally agree that this should be put in the changelog, since the difference is not necessarily, immediately obvious, but I don't think that there's actually a bug here or that there's a problem with ~.
Comment #4 by ricejm01 — 2016-11-25T02:51:59Z
I actually have to complain about this whole thing. In C/C++, Java, C# and virtually every programming language based on the C syntax allows for implicit string concatenation. Allowing this to be depreciated you are making the D language harder to use. the rationale for the reasoning for why all of sudden we have to start using ~ to break up long strings so that the code is readable is quite frankly Bullshit! Case in point. this: string[] names = [ "Anna", "Michael" "Emma", "David" ]; // The content of arr is [ "Anna", "MichaelEmma", "David" ] Vs. string aReallyLongString = "this is a really long string that will" "not look good on the page if it put on" " a single line"; There is a huge difference between Array assignment and single string assignment. How about you revert the changes in 2.072 and re-implement to fix the actual problem, which solely with array assignment and look for '[' ']' characters to not allow implicit string concatenation. At this point 2.072 is not usable, and I would appreciate not having to fix ~20,000 lines of code to add '~' everywhere. Not to mention you just broke almost every dub library I use as well.
Comment #5 by robert.schadek — 2024-12-15T15:23:56Z
THIS ISSUE HAS BEEN MOVED TO GITHUB https://github.com/dlang/dlang.org/issues/3960 DO NOT COMMENT HERE ANYMORE, NOBODY WILL SEE IT, THIS ISSUE HAS BEEN MOVED TO GITHUB