2014-04-03 19:55:06 -05:00
|
|
|
// Copyright 2014 The Rust Project Developers. See the COPYRIGHT
|
log: Introduce liblog, the old std::logging
This commit moves all logging out of the standard library into an external
crate. This crate is the new crate which is responsible for all logging macros
and logging implementation. A few reasons for this change are:
* The crate map has always been a bit of a code smell among rust programs. It
has difficulty being loaded on almost all platforms, and it's used almost
exclusively for logging and only logging. Removing the crate map is one of the
end goals of this movement.
* The compiler has a fair bit of special support for logging. It has the
__log_level() expression as well as generating a global word per module
specifying the log level. This is unfairly favoring the built-in logging
system, and is much better done purely in libraries instead of the compiler
itself.
* Initialization of logging is much easier to do if there is no reliance on a
magical crate map being available to set module log levels.
* If the logging library can be written outside of the standard library, there's
no reason that it shouldn't be. It's likely that we're not going to build the
highest quality logging library of all time, so third-party libraries should
be able to provide just as high-quality logging systems as the default one
provided in the rust distribution.
With a migration such as this, the change does not come for free. There are some
subtle changes in the behavior of liblog vs the previous logging macros:
* The core change of this migration is that there is no longer a physical
log-level per module. This concept is still emulated (it is quite useful), but
there is now only a global log level, not a local one. This global log level
is a reflection of the maximum of all log levels specified. The previously
generated logging code looked like:
if specified_level <= __module_log_level() {
println!(...)
}
The newly generated code looks like:
if specified_level <= ::log::LOG_LEVEL {
if ::log::module_enabled(module_path!()) {
println!(...)
}
}
Notably, the first layer of checking is still intended to be "super fast" in
that it's just a load of a global word and a compare. The second layer of
checking is executed to determine if the current module does indeed have
logging turned on.
This means that if any module has a debug log level turned on, all modules
with debug log levels get a little bit slower (they all do more expensive
dynamic checks to determine if they're turned on or not).
Semantically, this migration brings no change in this respect, but
runtime-wise, this will have a perf impact on some code.
* A `RUST_LOG=::help` directive will no longer print out a list of all modules
that can be logged. This is because the crate map will no longer specify the
log levels of all modules, so the list of modules is not known. Additionally,
warnings can no longer be provided if a malformed logging directive was
supplied.
The new "hello world" for logging looks like:
#[phase(syntax, link)]
extern crate log;
fn main() {
debug!("Hello, world!");
}
2014-03-09 00:11:44 -06:00
|
|
|
// file at the top-level directory of this distribution and at
|
|
|
|
// http://rust-lang.org/COPYRIGHT.
|
|
|
|
//
|
|
|
|
// Licensed under the Apache License, Version 2.0 <LICENSE-APACHE or
|
|
|
|
// http://www.apache.org/licenses/LICENSE-2.0> or the MIT license
|
|
|
|
// <LICENSE-MIT or http://opensource.org/licenses/MIT>, at your
|
|
|
|
// option. This file may not be copied, modified, or distributed
|
|
|
|
// except according to those terms.
|
|
|
|
|
|
|
|
/*!
|
|
|
|
|
|
|
|
Utilities for program-wide and customizable logging
|
|
|
|
|
2014-03-13 01:34:31 -05:00
|
|
|
## Example
|
|
|
|
|
|
|
|
```
|
2014-04-03 19:55:06 -05:00
|
|
|
#![feature(phase)]
|
2014-03-13 01:34:31 -05:00
|
|
|
#[phase(syntax, link)] extern crate log;
|
|
|
|
|
|
|
|
fn main() {
|
|
|
|
debug!("this is a debug {}", "message");
|
|
|
|
error!("this is printed by default");
|
|
|
|
|
|
|
|
if log_enabled!(log::INFO) {
|
|
|
|
let x = 3 * 4; // expensive computation
|
|
|
|
info!("the answer was: {}", x);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
```
|
|
|
|
|
|
|
|
## Logging Macros
|
log: Introduce liblog, the old std::logging
This commit moves all logging out of the standard library into an external
crate. This crate is the new crate which is responsible for all logging macros
and logging implementation. A few reasons for this change are:
* The crate map has always been a bit of a code smell among rust programs. It
has difficulty being loaded on almost all platforms, and it's used almost
exclusively for logging and only logging. Removing the crate map is one of the
end goals of this movement.
* The compiler has a fair bit of special support for logging. It has the
__log_level() expression as well as generating a global word per module
specifying the log level. This is unfairly favoring the built-in logging
system, and is much better done purely in libraries instead of the compiler
itself.
* Initialization of logging is much easier to do if there is no reliance on a
magical crate map being available to set module log levels.
* If the logging library can be written outside of the standard library, there's
no reason that it shouldn't be. It's likely that we're not going to build the
highest quality logging library of all time, so third-party libraries should
be able to provide just as high-quality logging systems as the default one
provided in the rust distribution.
With a migration such as this, the change does not come for free. There are some
subtle changes in the behavior of liblog vs the previous logging macros:
* The core change of this migration is that there is no longer a physical
log-level per module. This concept is still emulated (it is quite useful), but
there is now only a global log level, not a local one. This global log level
is a reflection of the maximum of all log levels specified. The previously
generated logging code looked like:
if specified_level <= __module_log_level() {
println!(...)
}
The newly generated code looks like:
if specified_level <= ::log::LOG_LEVEL {
if ::log::module_enabled(module_path!()) {
println!(...)
}
}
Notably, the first layer of checking is still intended to be "super fast" in
that it's just a load of a global word and a compare. The second layer of
checking is executed to determine if the current module does indeed have
logging turned on.
This means that if any module has a debug log level turned on, all modules
with debug log levels get a little bit slower (they all do more expensive
dynamic checks to determine if they're turned on or not).
Semantically, this migration brings no change in this respect, but
runtime-wise, this will have a perf impact on some code.
* A `RUST_LOG=::help` directive will no longer print out a list of all modules
that can be logged. This is because the crate map will no longer specify the
log levels of all modules, so the list of modules is not known. Additionally,
warnings can no longer be provided if a malformed logging directive was
supplied.
The new "hello world" for logging looks like:
#[phase(syntax, link)]
extern crate log;
fn main() {
debug!("Hello, world!");
}
2014-03-09 00:11:44 -06:00
|
|
|
|
|
|
|
There are five macros that the logging subsystem uses:
|
|
|
|
|
|
|
|
* `log!(level, ...)` - the generic logging macro, takes a level as a u32 and any
|
|
|
|
related `format!` arguments
|
|
|
|
* `debug!(...)` - a macro hard-wired to the log level of `DEBUG`
|
|
|
|
* `info!(...)` - a macro hard-wired to the log level of `INFO`
|
|
|
|
* `warn!(...)` - a macro hard-wired to the log level of `WARN`
|
|
|
|
* `error!(...)` - a macro hard-wired to the log level of `ERROR`
|
|
|
|
|
2014-04-11 09:28:03 -05:00
|
|
|
All of these macros use the same style of syntax as the `format!` syntax
|
log: Introduce liblog, the old std::logging
This commit moves all logging out of the standard library into an external
crate. This crate is the new crate which is responsible for all logging macros
and logging implementation. A few reasons for this change are:
* The crate map has always been a bit of a code smell among rust programs. It
has difficulty being loaded on almost all platforms, and it's used almost
exclusively for logging and only logging. Removing the crate map is one of the
end goals of this movement.
* The compiler has a fair bit of special support for logging. It has the
__log_level() expression as well as generating a global word per module
specifying the log level. This is unfairly favoring the built-in logging
system, and is much better done purely in libraries instead of the compiler
itself.
* Initialization of logging is much easier to do if there is no reliance on a
magical crate map being available to set module log levels.
* If the logging library can be written outside of the standard library, there's
no reason that it shouldn't be. It's likely that we're not going to build the
highest quality logging library of all time, so third-party libraries should
be able to provide just as high-quality logging systems as the default one
provided in the rust distribution.
With a migration such as this, the change does not come for free. There are some
subtle changes in the behavior of liblog vs the previous logging macros:
* The core change of this migration is that there is no longer a physical
log-level per module. This concept is still emulated (it is quite useful), but
there is now only a global log level, not a local one. This global log level
is a reflection of the maximum of all log levels specified. The previously
generated logging code looked like:
if specified_level <= __module_log_level() {
println!(...)
}
The newly generated code looks like:
if specified_level <= ::log::LOG_LEVEL {
if ::log::module_enabled(module_path!()) {
println!(...)
}
}
Notably, the first layer of checking is still intended to be "super fast" in
that it's just a load of a global word and a compare. The second layer of
checking is executed to determine if the current module does indeed have
logging turned on.
This means that if any module has a debug log level turned on, all modules
with debug log levels get a little bit slower (they all do more expensive
dynamic checks to determine if they're turned on or not).
Semantically, this migration brings no change in this respect, but
runtime-wise, this will have a perf impact on some code.
* A `RUST_LOG=::help` directive will no longer print out a list of all modules
that can be logged. This is because the crate map will no longer specify the
log levels of all modules, so the list of modules is not known. Additionally,
warnings can no longer be provided if a malformed logging directive was
supplied.
The new "hello world" for logging looks like:
#[phase(syntax, link)]
extern crate log;
fn main() {
debug!("Hello, world!");
}
2014-03-09 00:11:44 -06:00
|
|
|
extension. Details about the syntax can be found in the documentation of
|
|
|
|
`std::fmt` along with the Rust tutorial/manual.
|
|
|
|
|
|
|
|
If you want to check at runtime if a given logging level is enabled (e.g. if the
|
2014-04-11 09:28:03 -05:00
|
|
|
information you would want to log is expensive to produce), you can use the
|
log: Introduce liblog, the old std::logging
This commit moves all logging out of the standard library into an external
crate. This crate is the new crate which is responsible for all logging macros
and logging implementation. A few reasons for this change are:
* The crate map has always been a bit of a code smell among rust programs. It
has difficulty being loaded on almost all platforms, and it's used almost
exclusively for logging and only logging. Removing the crate map is one of the
end goals of this movement.
* The compiler has a fair bit of special support for logging. It has the
__log_level() expression as well as generating a global word per module
specifying the log level. This is unfairly favoring the built-in logging
system, and is much better done purely in libraries instead of the compiler
itself.
* Initialization of logging is much easier to do if there is no reliance on a
magical crate map being available to set module log levels.
* If the logging library can be written outside of the standard library, there's
no reason that it shouldn't be. It's likely that we're not going to build the
highest quality logging library of all time, so third-party libraries should
be able to provide just as high-quality logging systems as the default one
provided in the rust distribution.
With a migration such as this, the change does not come for free. There are some
subtle changes in the behavior of liblog vs the previous logging macros:
* The core change of this migration is that there is no longer a physical
log-level per module. This concept is still emulated (it is quite useful), but
there is now only a global log level, not a local one. This global log level
is a reflection of the maximum of all log levels specified. The previously
generated logging code looked like:
if specified_level <= __module_log_level() {
println!(...)
}
The newly generated code looks like:
if specified_level <= ::log::LOG_LEVEL {
if ::log::module_enabled(module_path!()) {
println!(...)
}
}
Notably, the first layer of checking is still intended to be "super fast" in
that it's just a load of a global word and a compare. The second layer of
checking is executed to determine if the current module does indeed have
logging turned on.
This means that if any module has a debug log level turned on, all modules
with debug log levels get a little bit slower (they all do more expensive
dynamic checks to determine if they're turned on or not).
Semantically, this migration brings no change in this respect, but
runtime-wise, this will have a perf impact on some code.
* A `RUST_LOG=::help` directive will no longer print out a list of all modules
that can be logged. This is because the crate map will no longer specify the
log levels of all modules, so the list of modules is not known. Additionally,
warnings can no longer be provided if a malformed logging directive was
supplied.
The new "hello world" for logging looks like:
#[phase(syntax, link)]
extern crate log;
fn main() {
debug!("Hello, world!");
}
2014-03-09 00:11:44 -06:00
|
|
|
following macro:
|
|
|
|
|
|
|
|
* `log_enabled!(level)` - returns true if logging of the given level is enabled
|
|
|
|
|
|
|
|
## Enabling logging
|
|
|
|
|
|
|
|
Log levels are controlled on a per-module basis, and by default all logging is
|
|
|
|
disabled except for `error!` (a log level of 1). Logging is controlled via the
|
|
|
|
`RUST_LOG` environment variable. The value of this environment variable is a
|
|
|
|
comma-separated list of logging directives. A logging directive is of the form:
|
|
|
|
|
|
|
|
```notrust
|
|
|
|
path::to::module=log_level
|
|
|
|
```
|
|
|
|
|
|
|
|
The path to the module is rooted in the name of the crate it was compiled for,
|
|
|
|
so if your program is contained in a file `hello.rs`, for example, to turn on
|
2014-04-11 09:28:03 -05:00
|
|
|
logging for this file you would use a value of `RUST_LOG=hello`.
|
log: Introduce liblog, the old std::logging
This commit moves all logging out of the standard library into an external
crate. This crate is the new crate which is responsible for all logging macros
and logging implementation. A few reasons for this change are:
* The crate map has always been a bit of a code smell among rust programs. It
has difficulty being loaded on almost all platforms, and it's used almost
exclusively for logging and only logging. Removing the crate map is one of the
end goals of this movement.
* The compiler has a fair bit of special support for logging. It has the
__log_level() expression as well as generating a global word per module
specifying the log level. This is unfairly favoring the built-in logging
system, and is much better done purely in libraries instead of the compiler
itself.
* Initialization of logging is much easier to do if there is no reliance on a
magical crate map being available to set module log levels.
* If the logging library can be written outside of the standard library, there's
no reason that it shouldn't be. It's likely that we're not going to build the
highest quality logging library of all time, so third-party libraries should
be able to provide just as high-quality logging systems as the default one
provided in the rust distribution.
With a migration such as this, the change does not come for free. There are some
subtle changes in the behavior of liblog vs the previous logging macros:
* The core change of this migration is that there is no longer a physical
log-level per module. This concept is still emulated (it is quite useful), but
there is now only a global log level, not a local one. This global log level
is a reflection of the maximum of all log levels specified. The previously
generated logging code looked like:
if specified_level <= __module_log_level() {
println!(...)
}
The newly generated code looks like:
if specified_level <= ::log::LOG_LEVEL {
if ::log::module_enabled(module_path!()) {
println!(...)
}
}
Notably, the first layer of checking is still intended to be "super fast" in
that it's just a load of a global word and a compare. The second layer of
checking is executed to determine if the current module does indeed have
logging turned on.
This means that if any module has a debug log level turned on, all modules
with debug log levels get a little bit slower (they all do more expensive
dynamic checks to determine if they're turned on or not).
Semantically, this migration brings no change in this respect, but
runtime-wise, this will have a perf impact on some code.
* A `RUST_LOG=::help` directive will no longer print out a list of all modules
that can be logged. This is because the crate map will no longer specify the
log levels of all modules, so the list of modules is not known. Additionally,
warnings can no longer be provided if a malformed logging directive was
supplied.
The new "hello world" for logging looks like:
#[phase(syntax, link)]
extern crate log;
fn main() {
debug!("Hello, world!");
}
2014-03-09 00:11:44 -06:00
|
|
|
Furthermore, this path is a prefix-search, so all modules nested in the
|
|
|
|
specified module will also have logging enabled.
|
|
|
|
|
|
|
|
The actual `log_level` is optional to specify. If omitted, all logging will be
|
|
|
|
enabled. If specified, the it must be either a numeric in the range of 1-255, or
|
|
|
|
it must be one of the strings `debug`, `error`, `info`, or `warn`. If a numeric
|
|
|
|
is specified, then all logging less than or equal to that numeral is enabled.
|
|
|
|
For example, if logging level 3 is active, error, warn, and info logs will be
|
|
|
|
printed, but debug will be omitted.
|
|
|
|
|
|
|
|
As the log level for a module is optional, the module to enable logging for is
|
|
|
|
also optional. If only a `log_level` is provided, then the global log level for
|
|
|
|
all modules is set to this value.
|
|
|
|
|
|
|
|
Some examples of valid values of `RUST_LOG` are:
|
|
|
|
|
|
|
|
```notrust
|
|
|
|
hello // turns on all logging for the 'hello' module
|
|
|
|
info // turns on all info logging
|
|
|
|
hello=debug // turns on debug logging for 'hello'
|
|
|
|
hello=3 // turns on info logging for 'hello'
|
|
|
|
hello,std::option // turns on hello, and std's option logging
|
|
|
|
error,hello=warn // turn on global error logging and also warn for hello
|
|
|
|
```
|
|
|
|
|
|
|
|
## Performance and Side Effects
|
|
|
|
|
|
|
|
Each of these macros will expand to code similar to:
|
|
|
|
|
|
|
|
```rust,ignore
|
|
|
|
if log_level <= my_module_log_level() {
|
|
|
|
::log::log(log_level, format!(...));
|
|
|
|
}
|
|
|
|
```
|
|
|
|
|
|
|
|
What this means is that each of these macros are very cheap at runtime if
|
|
|
|
they're turned off (just a load and an integer comparison). This also means that
|
|
|
|
if logging is disabled, none of the components of the log will be executed.
|
|
|
|
|
|
|
|
*/
|
|
|
|
|
2014-04-03 18:28:46 -05:00
|
|
|
#![crate_id = "log#0.11-pre"]
|
2014-03-21 20:05:05 -05:00
|
|
|
#![license = "MIT/ASL2"]
|
|
|
|
#![crate_type = "rlib"]
|
|
|
|
#![crate_type = "dylib"]
|
|
|
|
#![doc(html_logo_url = "http://www.rust-lang.org/logos/rust-logo-128x128-blk-v2.png",
|
|
|
|
html_favicon_url = "http://www.rust-lang.org/favicon.ico",
|
|
|
|
html_root_url = "http://static.rust-lang.org/doc/master")]
|
|
|
|
|
|
|
|
#![feature(macro_rules)]
|
|
|
|
#![deny(missing_doc, deprecated_owned_vector)]
|
log: Introduce liblog, the old std::logging
This commit moves all logging out of the standard library into an external
crate. This crate is the new crate which is responsible for all logging macros
and logging implementation. A few reasons for this change are:
* The crate map has always been a bit of a code smell among rust programs. It
has difficulty being loaded on almost all platforms, and it's used almost
exclusively for logging and only logging. Removing the crate map is one of the
end goals of this movement.
* The compiler has a fair bit of special support for logging. It has the
__log_level() expression as well as generating a global word per module
specifying the log level. This is unfairly favoring the built-in logging
system, and is much better done purely in libraries instead of the compiler
itself.
* Initialization of logging is much easier to do if there is no reliance on a
magical crate map being available to set module log levels.
* If the logging library can be written outside of the standard library, there's
no reason that it shouldn't be. It's likely that we're not going to build the
highest quality logging library of all time, so third-party libraries should
be able to provide just as high-quality logging systems as the default one
provided in the rust distribution.
With a migration such as this, the change does not come for free. There are some
subtle changes in the behavior of liblog vs the previous logging macros:
* The core change of this migration is that there is no longer a physical
log-level per module. This concept is still emulated (it is quite useful), but
there is now only a global log level, not a local one. This global log level
is a reflection of the maximum of all log levels specified. The previously
generated logging code looked like:
if specified_level <= __module_log_level() {
println!(...)
}
The newly generated code looks like:
if specified_level <= ::log::LOG_LEVEL {
if ::log::module_enabled(module_path!()) {
println!(...)
}
}
Notably, the first layer of checking is still intended to be "super fast" in
that it's just a load of a global word and a compare. The second layer of
checking is executed to determine if the current module does indeed have
logging turned on.
This means that if any module has a debug log level turned on, all modules
with debug log levels get a little bit slower (they all do more expensive
dynamic checks to determine if they're turned on or not).
Semantically, this migration brings no change in this respect, but
runtime-wise, this will have a perf impact on some code.
* A `RUST_LOG=::help` directive will no longer print out a list of all modules
that can be logged. This is because the crate map will no longer specify the
log levels of all modules, so the list of modules is not known. Additionally,
warnings can no longer be provided if a malformed logging directive was
supplied.
The new "hello world" for logging looks like:
#[phase(syntax, link)]
extern crate log;
fn main() {
debug!("Hello, world!");
}
2014-03-09 00:11:44 -06:00
|
|
|
|
|
|
|
extern crate sync;
|
|
|
|
|
|
|
|
use std::cast;
|
|
|
|
use std::fmt;
|
|
|
|
use std::io::LineBufferedWriter;
|
|
|
|
use std::io;
|
|
|
|
use std::local_data;
|
|
|
|
use std::os;
|
|
|
|
use std::rt;
|
2014-03-08 17:11:52 -06:00
|
|
|
use std::slice;
|
log: Introduce liblog, the old std::logging
This commit moves all logging out of the standard library into an external
crate. This crate is the new crate which is responsible for all logging macros
and logging implementation. A few reasons for this change are:
* The crate map has always been a bit of a code smell among rust programs. It
has difficulty being loaded on almost all platforms, and it's used almost
exclusively for logging and only logging. Removing the crate map is one of the
end goals of this movement.
* The compiler has a fair bit of special support for logging. It has the
__log_level() expression as well as generating a global word per module
specifying the log level. This is unfairly favoring the built-in logging
system, and is much better done purely in libraries instead of the compiler
itself.
* Initialization of logging is much easier to do if there is no reliance on a
magical crate map being available to set module log levels.
* If the logging library can be written outside of the standard library, there's
no reason that it shouldn't be. It's likely that we're not going to build the
highest quality logging library of all time, so third-party libraries should
be able to provide just as high-quality logging systems as the default one
provided in the rust distribution.
With a migration such as this, the change does not come for free. There are some
subtle changes in the behavior of liblog vs the previous logging macros:
* The core change of this migration is that there is no longer a physical
log-level per module. This concept is still emulated (it is quite useful), but
there is now only a global log level, not a local one. This global log level
is a reflection of the maximum of all log levels specified. The previously
generated logging code looked like:
if specified_level <= __module_log_level() {
println!(...)
}
The newly generated code looks like:
if specified_level <= ::log::LOG_LEVEL {
if ::log::module_enabled(module_path!()) {
println!(...)
}
}
Notably, the first layer of checking is still intended to be "super fast" in
that it's just a load of a global word and a compare. The second layer of
checking is executed to determine if the current module does indeed have
logging turned on.
This means that if any module has a debug log level turned on, all modules
with debug log levels get a little bit slower (they all do more expensive
dynamic checks to determine if they're turned on or not).
Semantically, this migration brings no change in this respect, but
runtime-wise, this will have a perf impact on some code.
* A `RUST_LOG=::help` directive will no longer print out a list of all modules
that can be logged. This is because the crate map will no longer specify the
log levels of all modules, so the list of modules is not known. Additionally,
warnings can no longer be provided if a malformed logging directive was
supplied.
The new "hello world" for logging looks like:
#[phase(syntax, link)]
extern crate log;
fn main() {
debug!("Hello, world!");
}
2014-03-09 00:11:44 -06:00
|
|
|
|
|
|
|
use sync::one::{Once, ONCE_INIT};
|
|
|
|
|
|
|
|
pub mod macros;
|
|
|
|
mod directive;
|
|
|
|
|
|
|
|
/// Maximum logging level of a module that can be specified. Common logging
|
|
|
|
/// levels are found in the DEBUG/INFO/WARN/ERROR constants.
|
|
|
|
pub static MAX_LOG_LEVEL: u32 = 255;
|
|
|
|
|
|
|
|
/// The default logging level of a crate if no other is specified.
|
|
|
|
static DEFAULT_LOG_LEVEL: u32 = 1;
|
|
|
|
|
|
|
|
/// An unsafe constant that is the maximum logging level of any module
|
|
|
|
/// specified. This is the first line of defense to determining whether a
|
|
|
|
/// logging statement should be run.
|
|
|
|
static mut LOG_LEVEL: u32 = MAX_LOG_LEVEL;
|
|
|
|
|
|
|
|
static mut DIRECTIVES: *Vec<directive::LogDirective> =
|
|
|
|
0 as *Vec<directive::LogDirective>;
|
|
|
|
|
|
|
|
/// Debug log level
|
|
|
|
pub static DEBUG: u32 = 4;
|
|
|
|
/// Info log level
|
|
|
|
pub static INFO: u32 = 3;
|
|
|
|
/// Warn log level
|
|
|
|
pub static WARN: u32 = 2;
|
|
|
|
/// Error log level
|
|
|
|
pub static ERROR: u32 = 1;
|
|
|
|
|
2014-03-08 20:21:49 -06:00
|
|
|
local_data_key!(local_logger: ~Logger:Send)
|
log: Introduce liblog, the old std::logging
This commit moves all logging out of the standard library into an external
crate. This crate is the new crate which is responsible for all logging macros
and logging implementation. A few reasons for this change are:
* The crate map has always been a bit of a code smell among rust programs. It
has difficulty being loaded on almost all platforms, and it's used almost
exclusively for logging and only logging. Removing the crate map is one of the
end goals of this movement.
* The compiler has a fair bit of special support for logging. It has the
__log_level() expression as well as generating a global word per module
specifying the log level. This is unfairly favoring the built-in logging
system, and is much better done purely in libraries instead of the compiler
itself.
* Initialization of logging is much easier to do if there is no reliance on a
magical crate map being available to set module log levels.
* If the logging library can be written outside of the standard library, there's
no reason that it shouldn't be. It's likely that we're not going to build the
highest quality logging library of all time, so third-party libraries should
be able to provide just as high-quality logging systems as the default one
provided in the rust distribution.
With a migration such as this, the change does not come for free. There are some
subtle changes in the behavior of liblog vs the previous logging macros:
* The core change of this migration is that there is no longer a physical
log-level per module. This concept is still emulated (it is quite useful), but
there is now only a global log level, not a local one. This global log level
is a reflection of the maximum of all log levels specified. The previously
generated logging code looked like:
if specified_level <= __module_log_level() {
println!(...)
}
The newly generated code looks like:
if specified_level <= ::log::LOG_LEVEL {
if ::log::module_enabled(module_path!()) {
println!(...)
}
}
Notably, the first layer of checking is still intended to be "super fast" in
that it's just a load of a global word and a compare. The second layer of
checking is executed to determine if the current module does indeed have
logging turned on.
This means that if any module has a debug log level turned on, all modules
with debug log levels get a little bit slower (they all do more expensive
dynamic checks to determine if they're turned on or not).
Semantically, this migration brings no change in this respect, but
runtime-wise, this will have a perf impact on some code.
* A `RUST_LOG=::help` directive will no longer print out a list of all modules
that can be logged. This is because the crate map will no longer specify the
log levels of all modules, so the list of modules is not known. Additionally,
warnings can no longer be provided if a malformed logging directive was
supplied.
The new "hello world" for logging looks like:
#[phase(syntax, link)]
extern crate log;
fn main() {
debug!("Hello, world!");
}
2014-03-09 00:11:44 -06:00
|
|
|
|
|
|
|
/// A trait used to represent an interface to a task-local logger. Each task
|
|
|
|
/// can have its own custom logger which can respond to logging messages
|
|
|
|
/// however it likes.
|
|
|
|
pub trait Logger {
|
|
|
|
/// Logs a single message described by the `args` structure. The level is
|
|
|
|
/// provided in case you want to do things like color the message, etc.
|
|
|
|
fn log(&mut self, level: u32, args: &fmt::Arguments);
|
|
|
|
}
|
|
|
|
|
|
|
|
struct DefaultLogger {
|
|
|
|
handle: LineBufferedWriter<io::stdio::StdWriter>,
|
|
|
|
}
|
|
|
|
|
|
|
|
impl Logger for DefaultLogger {
|
|
|
|
// by default, just ignore the level
|
|
|
|
fn log(&mut self, _level: u32, args: &fmt::Arguments) {
|
|
|
|
match fmt::writeln(&mut self.handle, args) {
|
|
|
|
Err(e) => fail!("failed to log: {}", e),
|
|
|
|
Ok(()) => {}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
impl Drop for DefaultLogger {
|
|
|
|
fn drop(&mut self) {
|
|
|
|
// FIXME(#12628): is failure the right thing to do?
|
|
|
|
match self.handle.flush() {
|
|
|
|
Err(e) => fail!("failed to flush a logger: {}", e),
|
|
|
|
Ok(()) => {}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/// This function is called directly by the compiler when using the logging
|
|
|
|
/// macros. This function does not take into account whether the log level
|
|
|
|
/// specified is active or not, it will always log something if this method is
|
|
|
|
/// called.
|
|
|
|
///
|
|
|
|
/// It is not recommended to call this function directly, rather it should be
|
|
|
|
/// invoked through the logging family of macros.
|
|
|
|
pub fn log(level: u32, args: &fmt::Arguments) {
|
|
|
|
// Completely remove the local logger from TLS in case anyone attempts to
|
|
|
|
// frob the slot while we're doing the logging. This will destroy any logger
|
|
|
|
// set during logging.
|
|
|
|
let mut logger = local_data::pop(local_logger).unwrap_or_else(|| {
|
2014-03-08 20:21:49 -06:00
|
|
|
~DefaultLogger { handle: io::stderr() } as ~Logger:Send
|
log: Introduce liblog, the old std::logging
This commit moves all logging out of the standard library into an external
crate. This crate is the new crate which is responsible for all logging macros
and logging implementation. A few reasons for this change are:
* The crate map has always been a bit of a code smell among rust programs. It
has difficulty being loaded on almost all platforms, and it's used almost
exclusively for logging and only logging. Removing the crate map is one of the
end goals of this movement.
* The compiler has a fair bit of special support for logging. It has the
__log_level() expression as well as generating a global word per module
specifying the log level. This is unfairly favoring the built-in logging
system, and is much better done purely in libraries instead of the compiler
itself.
* Initialization of logging is much easier to do if there is no reliance on a
magical crate map being available to set module log levels.
* If the logging library can be written outside of the standard library, there's
no reason that it shouldn't be. It's likely that we're not going to build the
highest quality logging library of all time, so third-party libraries should
be able to provide just as high-quality logging systems as the default one
provided in the rust distribution.
With a migration such as this, the change does not come for free. There are some
subtle changes in the behavior of liblog vs the previous logging macros:
* The core change of this migration is that there is no longer a physical
log-level per module. This concept is still emulated (it is quite useful), but
there is now only a global log level, not a local one. This global log level
is a reflection of the maximum of all log levels specified. The previously
generated logging code looked like:
if specified_level <= __module_log_level() {
println!(...)
}
The newly generated code looks like:
if specified_level <= ::log::LOG_LEVEL {
if ::log::module_enabled(module_path!()) {
println!(...)
}
}
Notably, the first layer of checking is still intended to be "super fast" in
that it's just a load of a global word and a compare. The second layer of
checking is executed to determine if the current module does indeed have
logging turned on.
This means that if any module has a debug log level turned on, all modules
with debug log levels get a little bit slower (they all do more expensive
dynamic checks to determine if they're turned on or not).
Semantically, this migration brings no change in this respect, but
runtime-wise, this will have a perf impact on some code.
* A `RUST_LOG=::help` directive will no longer print out a list of all modules
that can be logged. This is because the crate map will no longer specify the
log levels of all modules, so the list of modules is not known. Additionally,
warnings can no longer be provided if a malformed logging directive was
supplied.
The new "hello world" for logging looks like:
#[phase(syntax, link)]
extern crate log;
fn main() {
debug!("Hello, world!");
}
2014-03-09 00:11:44 -06:00
|
|
|
});
|
|
|
|
logger.log(level, args);
|
|
|
|
local_data::set(local_logger, logger);
|
|
|
|
}
|
|
|
|
|
|
|
|
/// Getter for the global log level. This is a function so that it can be called
|
|
|
|
/// safely
|
|
|
|
#[doc(hidden)]
|
|
|
|
#[inline(always)]
|
|
|
|
pub fn log_level() -> u32 { unsafe { LOG_LEVEL } }
|
|
|
|
|
|
|
|
/// Replaces the task-local logger with the specified logger, returning the old
|
|
|
|
/// logger.
|
2014-03-08 20:21:49 -06:00
|
|
|
pub fn set_logger(logger: ~Logger:Send) -> Option<~Logger:Send> {
|
log: Introduce liblog, the old std::logging
This commit moves all logging out of the standard library into an external
crate. This crate is the new crate which is responsible for all logging macros
and logging implementation. A few reasons for this change are:
* The crate map has always been a bit of a code smell among rust programs. It
has difficulty being loaded on almost all platforms, and it's used almost
exclusively for logging and only logging. Removing the crate map is one of the
end goals of this movement.
* The compiler has a fair bit of special support for logging. It has the
__log_level() expression as well as generating a global word per module
specifying the log level. This is unfairly favoring the built-in logging
system, and is much better done purely in libraries instead of the compiler
itself.
* Initialization of logging is much easier to do if there is no reliance on a
magical crate map being available to set module log levels.
* If the logging library can be written outside of the standard library, there's
no reason that it shouldn't be. It's likely that we're not going to build the
highest quality logging library of all time, so third-party libraries should
be able to provide just as high-quality logging systems as the default one
provided in the rust distribution.
With a migration such as this, the change does not come for free. There are some
subtle changes in the behavior of liblog vs the previous logging macros:
* The core change of this migration is that there is no longer a physical
log-level per module. This concept is still emulated (it is quite useful), but
there is now only a global log level, not a local one. This global log level
is a reflection of the maximum of all log levels specified. The previously
generated logging code looked like:
if specified_level <= __module_log_level() {
println!(...)
}
The newly generated code looks like:
if specified_level <= ::log::LOG_LEVEL {
if ::log::module_enabled(module_path!()) {
println!(...)
}
}
Notably, the first layer of checking is still intended to be "super fast" in
that it's just a load of a global word and a compare. The second layer of
checking is executed to determine if the current module does indeed have
logging turned on.
This means that if any module has a debug log level turned on, all modules
with debug log levels get a little bit slower (they all do more expensive
dynamic checks to determine if they're turned on or not).
Semantically, this migration brings no change in this respect, but
runtime-wise, this will have a perf impact on some code.
* A `RUST_LOG=::help` directive will no longer print out a list of all modules
that can be logged. This is because the crate map will no longer specify the
log levels of all modules, so the list of modules is not known. Additionally,
warnings can no longer be provided if a malformed logging directive was
supplied.
The new "hello world" for logging looks like:
#[phase(syntax, link)]
extern crate log;
fn main() {
debug!("Hello, world!");
}
2014-03-09 00:11:44 -06:00
|
|
|
let prev = local_data::pop(local_logger);
|
|
|
|
local_data::set(local_logger, logger);
|
|
|
|
return prev;
|
|
|
|
}
|
|
|
|
|
|
|
|
/// Tests whether a given module's name is enabled for a particular level of
|
|
|
|
/// logging. This is the second layer of defense about determining whether a
|
|
|
|
/// module's log statement should be emitted or not.
|
|
|
|
#[doc(hidden)]
|
|
|
|
pub fn mod_enabled(level: u32, module: &str) -> bool {
|
|
|
|
static mut INIT: Once = ONCE_INIT;
|
|
|
|
unsafe { INIT.doit(init); }
|
|
|
|
|
|
|
|
// It's possible for many threads are in this function, only one of them
|
|
|
|
// will peform the global initialization, but all of them will need to check
|
|
|
|
// again to whether they should really be here or not. Hence, despite this
|
|
|
|
// check being expanded manually in the logging macro, this function checks
|
|
|
|
// the log level again.
|
|
|
|
if level > unsafe { LOG_LEVEL } { return false }
|
|
|
|
|
|
|
|
// This assertion should never get tripped unless we're in an at_exit
|
|
|
|
// handler after logging has been torn down and a logging attempt was made.
|
|
|
|
assert!(unsafe { !DIRECTIVES.is_null() });
|
|
|
|
|
|
|
|
enabled(level, module, unsafe { (*DIRECTIVES).iter() })
|
|
|
|
}
|
|
|
|
|
|
|
|
fn enabled(level: u32, module: &str,
|
2014-03-08 17:11:52 -06:00
|
|
|
iter: slice::Items<directive::LogDirective>) -> bool {
|
log: Introduce liblog, the old std::logging
This commit moves all logging out of the standard library into an external
crate. This crate is the new crate which is responsible for all logging macros
and logging implementation. A few reasons for this change are:
* The crate map has always been a bit of a code smell among rust programs. It
has difficulty being loaded on almost all platforms, and it's used almost
exclusively for logging and only logging. Removing the crate map is one of the
end goals of this movement.
* The compiler has a fair bit of special support for logging. It has the
__log_level() expression as well as generating a global word per module
specifying the log level. This is unfairly favoring the built-in logging
system, and is much better done purely in libraries instead of the compiler
itself.
* Initialization of logging is much easier to do if there is no reliance on a
magical crate map being available to set module log levels.
* If the logging library can be written outside of the standard library, there's
no reason that it shouldn't be. It's likely that we're not going to build the
highest quality logging library of all time, so third-party libraries should
be able to provide just as high-quality logging systems as the default one
provided in the rust distribution.
With a migration such as this, the change does not come for free. There are some
subtle changes in the behavior of liblog vs the previous logging macros:
* The core change of this migration is that there is no longer a physical
log-level per module. This concept is still emulated (it is quite useful), but
there is now only a global log level, not a local one. This global log level
is a reflection of the maximum of all log levels specified. The previously
generated logging code looked like:
if specified_level <= __module_log_level() {
println!(...)
}
The newly generated code looks like:
if specified_level <= ::log::LOG_LEVEL {
if ::log::module_enabled(module_path!()) {
println!(...)
}
}
Notably, the first layer of checking is still intended to be "super fast" in
that it's just a load of a global word and a compare. The second layer of
checking is executed to determine if the current module does indeed have
logging turned on.
This means that if any module has a debug log level turned on, all modules
with debug log levels get a little bit slower (they all do more expensive
dynamic checks to determine if they're turned on or not).
Semantically, this migration brings no change in this respect, but
runtime-wise, this will have a perf impact on some code.
* A `RUST_LOG=::help` directive will no longer print out a list of all modules
that can be logged. This is because the crate map will no longer specify the
log levels of all modules, so the list of modules is not known. Additionally,
warnings can no longer be provided if a malformed logging directive was
supplied.
The new "hello world" for logging looks like:
#[phase(syntax, link)]
extern crate log;
fn main() {
debug!("Hello, world!");
}
2014-03-09 00:11:44 -06:00
|
|
|
// Search for the longest match, the vector is assumed to be pre-sorted.
|
|
|
|
for directive in iter.rev() {
|
|
|
|
match directive.name {
|
|
|
|
Some(ref name) if !module.starts_with(*name) => {},
|
|
|
|
Some(..) | None => {
|
|
|
|
return level <= directive.level
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
level <= DEFAULT_LOG_LEVEL
|
|
|
|
}
|
|
|
|
|
|
|
|
/// Initialize logging for the current process.
|
|
|
|
///
|
|
|
|
/// This is not threadsafe at all, so initialization os performed through a
|
|
|
|
/// `Once` primitive (and this function is called from that primitive).
|
|
|
|
fn init() {
|
|
|
|
let mut directives = match os::getenv("RUST_LOG") {
|
|
|
|
Some(spec) => directive::parse_logging_spec(spec),
|
|
|
|
None => Vec::new(),
|
|
|
|
};
|
|
|
|
|
|
|
|
// Sort the provided directives by length of their name, this allows a
|
|
|
|
// little more efficient lookup at runtime.
|
|
|
|
directives.sort_by(|a, b| {
|
|
|
|
let alen = a.name.as_ref().map(|a| a.len()).unwrap_or(0);
|
|
|
|
let blen = b.name.as_ref().map(|b| b.len()).unwrap_or(0);
|
|
|
|
alen.cmp(&blen)
|
|
|
|
});
|
|
|
|
|
|
|
|
let max_level = {
|
|
|
|
let max = directives.iter().max_by(|d| d.level);
|
|
|
|
max.map(|d| d.level).unwrap_or(DEFAULT_LOG_LEVEL)
|
|
|
|
};
|
|
|
|
|
|
|
|
unsafe {
|
|
|
|
LOG_LEVEL = max_level;
|
|
|
|
|
|
|
|
assert!(DIRECTIVES.is_null());
|
|
|
|
DIRECTIVES = cast::transmute(~directives);
|
|
|
|
|
|
|
|
// Schedule the cleanup for this global for when the runtime exits.
|
|
|
|
rt::at_exit(proc() {
|
|
|
|
assert!(!DIRECTIVES.is_null());
|
|
|
|
let _directives: ~Vec<directive::LogDirective> =
|
|
|
|
cast::transmute(DIRECTIVES);
|
|
|
|
DIRECTIVES = 0 as *Vec<directive::LogDirective>;
|
|
|
|
});
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
#[cfg(test)]
|
|
|
|
mod tests {
|
|
|
|
use super::enabled;
|
|
|
|
use directive::LogDirective;
|
|
|
|
|
|
|
|
#[test]
|
|
|
|
fn match_full_path() {
|
|
|
|
let dirs = [LogDirective { name: Some(~"crate2"), level: 3 },
|
|
|
|
LogDirective { name: Some(~"crate1::mod1"), level: 2 }];
|
|
|
|
assert!(enabled(2, "crate1::mod1", dirs.iter()));
|
|
|
|
assert!(!enabled(3, "crate1::mod1", dirs.iter()));
|
|
|
|
assert!(enabled(3, "crate2", dirs.iter()));
|
|
|
|
assert!(!enabled(4, "crate2", dirs.iter()));
|
|
|
|
}
|
|
|
|
|
|
|
|
#[test]
|
|
|
|
fn no_match() {
|
|
|
|
let dirs = [LogDirective { name: Some(~"crate2"), level: 3 },
|
|
|
|
LogDirective { name: Some(~"crate1::mod1"), level: 2 }];
|
|
|
|
assert!(!enabled(2, "crate3", dirs.iter()));
|
|
|
|
}
|
|
|
|
|
|
|
|
#[test]
|
|
|
|
fn match_beginning() {
|
|
|
|
let dirs = [LogDirective { name: Some(~"crate2"), level: 3 },
|
|
|
|
LogDirective { name: Some(~"crate1::mod1"), level: 2 }];
|
|
|
|
assert!(enabled(3, "crate2::mod1", dirs.iter()));
|
|
|
|
}
|
|
|
|
|
|
|
|
#[test]
|
|
|
|
fn match_beginning_longest_match() {
|
|
|
|
let dirs = [LogDirective { name: Some(~"crate2"), level: 3 },
|
|
|
|
LogDirective { name: Some(~"crate2::mod"), level: 4 },
|
|
|
|
LogDirective { name: Some(~"crate1::mod1"), level: 2 }];
|
|
|
|
assert!(enabled(4, "crate2::mod1", dirs.iter()));
|
|
|
|
assert!(!enabled(4, "crate2", dirs.iter()));
|
|
|
|
}
|
|
|
|
|
|
|
|
#[test]
|
|
|
|
fn match_default() {
|
|
|
|
let dirs = [LogDirective { name: None, level: 3 },
|
|
|
|
LogDirective { name: Some(~"crate1::mod1"), level: 2 }];
|
|
|
|
assert!(enabled(2, "crate1::mod1", dirs.iter()));
|
|
|
|
assert!(enabled(3, "crate2::mod2", dirs.iter()));
|
|
|
|
}
|
|
|
|
|
|
|
|
#[test]
|
|
|
|
fn zero_level() {
|
|
|
|
let dirs = [LogDirective { name: None, level: 3 },
|
|
|
|
LogDirective { name: Some(~"crate1::mod1"), level: 0 }];
|
|
|
|
assert!(!enabled(1, "crate1::mod1", dirs.iter()));
|
|
|
|
assert!(enabled(3, "crate2::mod2", dirs.iter()));
|
|
|
|
}
|
|
|
|
}
|