The general consensus is that we want to move away from conditions for I/O, and I propose a two-step plan for doing so:
1. Warn about unused `Result` types. When all of I/O returns `Result`, it will require you inspect the return value for an error *only if* you have a result you want to look at. By default, for things like `write` returning `Result<(), Error>`, these will all go silently ignored. This lint will prevent blind ignorance of these return values, letting you know that there's something you should do about them.
2. Implement a `try!` macro:
```
macro_rules! try( ($e:expr) => (match $e { Ok(e) => e, Err(e) => return Err(e) }) )
```
With these two tools combined, I feel that we get almost all the benefits of conditions. The first step (the lint) is a sanity check that you're not ignoring return values at callsites. The second step is to provide a convenience method of returning early out of a sequence of computations. After thinking about this for awhile, I don't think that we need the so-called "do-notation" in the compiler itself because I think it's just *too* specialized. Additionally, the `try!` macro is super lightweight, easy to understand, and works almost everywhere. As soon as you want to do something more fancy, my answer is "use match".
Basically, with these two tools in action, I would be comfortable removing conditions. What do others think about this strategy?
----
This PR specifically implements the `unused_result` lint. I actually added two lints, `unused_result` and `unused_must_use`, and the first commit has the rationale for why `unused_result` is turned off by default.
I attempted to implement the lint in two steps. My first attempt was a
default-warn lint about *all* unused results. While this attempt did indeed find
many possible bugs, I felt that the false-positive rate was too high to be
turned on by default for all of Rust.
My second attempt was to make unused-result a default-allow lint, but allow
certain types to opt-in to the notion of "you must use this". For example, the
Result type is now flagged with #[must_use]. This lint about "must use" types is
warn by default (it's different from unused-result).
The unused_must_use lint had a 100% hit rate in the compiler, but there's not
that many places that return Result right now. I believe that this lint is a
crucial step towards moving away from conditions for I/O (because all I/O will
return Result by default). I'm worried that this lint is a little too specific
to Result itself, but I believe that the false positive rate for the
unused_result lint is too high to make it useful when turned on by default.
cc #7621.
See the commit message. I'm not sure if we should merge this now, or wait until we can write `Clone::clone(x)` which will directly solve the above issue with perfect error messages.
This unfortunately changes an error like
error: mismatched types: expected `&&NotClone` but found `&NotClone`
into
error: type `NotClone` does not implement any method in scope named `clone`
Non-exhaustive change list:
* `self` is now present in argument lists (modulo type-checking code I don't trust myself to refactor)
* methods have the same calling convention as bare functions (including the self argument)
* the env param is gone from all bare functions (and methods), only used by closures and `proc`s
* bare functions can only be coerced to closures and `proc`s if they are statically resolved, as they now require creating a wrapper specific to that function, to avoid indirect wrappers (equivalent to `impl<..Args, Ret> Fn<..Args, Ret> for fn(..Args) -> Ret`) that might not be optimizable by LLVM and don't work for `proc`s
* refactored some `trans::closure` code, leading to the removal of `trans::glue::make_free_glue` and `ty_opaque_closure_ptr`
I'd forgotten to update them when I changed this a while ago; it now displays error messages linked to the struct/variant field, rather than the `#[deriving(Trait)]` line, for all traits.
This also adds a very large number of autogenerated tests. I can easily remove/tone down that commit if necessary.
This was the original intention of the privacy of structs, and it was
erroneously implemented before. A pub struct will now have default-pub fields,
and a non-pub struct will have default-priv fields. This essentially brings
struct fields in line with enum variants in terms of inheriting visibility.
As usual, extraneous modifiers to visibility are disallowed depend on the case
that you're dealing with.
Closes#11522
They all have to go into a single module at the moment unfortunately.
Ideally, the logging macros would live in std::logging, condition! would
live in std::condition, format! in std::fmt, etc. However, this
introduces cyclic dependencies between those modules and the macros they
use which the current expansion system can't deal with. We may be able
to get around this by changing the expansion phase to a two-pass system
but that's for a later PR.
Closes#2247
cc #11763
A mutable and immutable borrow place some restrictions on what you can
with the variable until the borrow ends. This commit attempts to convey
to the user what those restrictions are. Also, if the original borrow is
a mutable borrow, the error message has been changed (more specifically,
i. "cannot borrow `x` as immutable because it is also borrowed as
mutable" and ii. "cannot borrow `x` as mutable more than once" have
been changed to "cannot borrow `x` because it is already borrowed as
mutable").
In addition, this adds a (custom) span note to communicate where the
original borrow ends.
Previously, they were treated like ~[] and &[] (which can have length
0), but fixed length vectors are fixed length, i.e. we know at compile
time if it's possible to have length zero (which is only for [T, .. 0]).
Fixes#11659.
This means that compilation continues for longer, and so we can see more
errors per compile. This is mildly more user-friendly because it stops
users having to run rustc n times to see n macro errors: just run it
once to see all of them.
Previously, they were treated like ~[] and &[] (which can have length
0), but fixed length vectors are fixed length, i.e. we know at compile
time if it's possible to have length zero (which is only for [T, .. 0]).
Fixes#11659.
For `use` statements, this means disallowing qualifiers when in functions and
disallowing `priv` outside of functions.
For `extern mod` statements, this means disallowing everything everywhere. It
may have been envisioned for `pub extern mod foo` to be a thing, but it
currently doesn't do anything (resolve doesn't pick it up), so better to err on
the side of forwards-compatibility and forbid it entirely for now.
Closes#9957
For `use` statements, this means disallowing qualifiers when in functions and
disallowing `priv` outside of functions.
For `extern mod` statements, this means disallowing everything everywhere. It
may have been envisioned for `pub extern mod foo` to be a thing, but it
currently doesn't do anything (resolve doesn't pick it up), so better to err on
the side of forwards-compatibility and forbid it entirely for now.
Closes#9957
Major changes:
- Define temporary scopes in a syntax-based way that basically defaults
to the innermost statement or conditional block, except for in
a `let` initializer, where we default to the innermost block. Rules
are documented in the code, but not in the manual (yet).
See new test run-pass/cleanup-value-scopes.rs for examples.
- Refactors Datum to better define cleanup roles.
- Refactor cleanup scopes to not be tied to basic blocks, permitting
us to have a very large number of scopes (one per AST node).
- Introduce nascent documentation in trans/doc.rs covering datums and
cleanup in a more comprehensive way.
r? @pcwalton
This means that compilation continues for longer, and so we can see more
errors per compile. This is mildly more user-friendly because it stops
users having to run rustc n times to see n macro errors: just run it
once to see all of them.
Major changes:
- Define temporary scopes in a syntax-based way that basically defaults
to the innermost statement or conditional block, except for in
a `let` initializer, where we default to the innermost block. Rules
are documented in the code, but not in the manual (yet).
See new test run-pass/cleanup-value-scopes.rs for examples.
- Refactors Datum to better define cleanup roles.
- Refactor cleanup scopes to not be tied to basic blocks, permitting
us to have a very large number of scopes (one per AST node).
- Introduce nascent documentation in trans/doc.rs covering datums and
cleanup in a more comprehensive way.
This is a patch for #8005, thanks @lfairy for the hint.
It seems like `block.expr` is None, if the last line of a function has a semi colon (= it ends with a statement).
@kmcallister does this error message cover the intended use cases?
I'm not sure about the message, the wording and the span could probably be improved.
Unsuffixed literals like 1 and 1.1, and free type parameters sometimes
have to be printed in error messages, which ended up with <V0>, <VI0>
and <VF0>. This change puts the words "generic" and "integer"/"float"
into the message so it's not a completely black box.
Dead code pass now explicitly checks for `#[allow(dead_code)]` and
`#[lang=".."]` attributes on items and marks them as live if they have
those attributes. The former is done so that if we want to suppress
warnings for a group of dead functions, we only have to annotate the
"root" of the call chain.