ruby-core@ruby-lang.org archive (unofficial mirror)
 help / color / mirror / Atom feed
From: akr@fsij.org
To: ruby-core@ruby-lang.org
Subject: [ruby-core:91501] [Ruby trunk Feature#15592] mode where "autoload" behaves like an immediate "require"
Date: Sat, 09 Feb 2019 13:36:37 +0000 (UTC)	[thread overview]
Message-ID: <redmine.journal-76761.20190209133635.04a1b02e829ca335@ruby-lang.org> (raw)
In-Reply-To: redmine.issue-15592.20190207065816@ruby-lang.org

Issue #15592 has been updated by akr (Akira Tanaka).


fxn (Xavier Noria) wrote:
> Zeitwerk has an interface to eager load similar to what it is being suggested, but it is able to do so easily because it controls the autoloads that it sets based solely in the file system.

I guess it needs some convention between constant names and filenames.
I agree it is good if such convention is available.  (convention over configuration)
Unfortunately, Ruby itself has no such reliable convention.

> A generic way to register and eager load as outlined above seems like a chicken and egg problem to me, because you have to evaluate the files in order to execute the autoloads, no? You need to eager load at least a bit to register the autoloads that trigger eager loading.

eager_load_autoloads method would load libraries until no autoload constant:

```
while there are at least one autoload constant
  load one of them (it may add more autoload constants)
end
```

This loop should terminate because there are only finite ruby library files.

Initial list of autoload constants are defined before user application before
eager_load_autoloads.

I think this can be implemented without much problems.


----------------------------------------
Feature #15592: mode where "autoload" behaves like an immediate "require"
https://bugs.ruby-lang.org/issues/15592#change-76761

* Author: akr (Akira Tanaka)
* Status: Open
* Priority: Normal
* Assignee: 
* Target version: 
----------------------------------------
How about a feature to switch "autoload" behavior to "require" immediately.

autoload is a feature for lazy loading.

matz dislikes autoload as [Feature #5653].
I heard that he dislikes class (and other) definitions at arbitrary timing.
I agree that eager loading is safer than lazy loading.

However, lazy loading realize shorter loading time and
it makes development cycle shorter.
It is more important for larger applications as Eregon said in
https://bugs.ruby-lang.org/issues/5653#note-39 .
It is especially important when library loading causes I/O (code generation from DB schema).

These two, safety of eager loading and easier development of lazy loading, conflicts.
But if we can distinguish production mode and development mode,
we can enjoy both benefits.

So, I propose a feature to select autoload behavior from two modes:
- autoload behaves as lazy loading as now in development mode
- autoload behaves as eager loading (immediately invokes "require") in production mode.

There are several idea to switch the mode:

- $AUTOLOAD_MODE = :eager or :lazy
- RubyVM.autoload_mode = :eager or :lazy
- ObjectSpace.autoload_mode = :eager or lazy

I'm not sure there is a good enough one in above list, though.



-- 
https://bugs.ruby-lang.org/

  parent reply	other threads:[~2019-02-09 13:36 UTC|newest]

Thread overview: 13+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
     [not found] <redmine.issue-15592.20190207065816@ruby-lang.org>
2019-02-07  6:58 ` [ruby-core:91454] [Ruby trunk Feature#15592] mode that "autoload" behaves "require" immediately akr
2019-02-07 14:35 ` [ruby-core:91472] " merch-redmine
2019-02-07 18:13 ` [ruby-core:91475] " eregontp
2019-02-07 20:21 ` [ruby-core:91477] " fxn
2019-02-07 20:45 ` [ruby-core:91478] " eregontp
2019-02-07 20:47 ` [ruby-core:91479] " eregontp
2019-02-07 20:49 ` [ruby-core:91480] " eregontp
2019-02-07 20:58 ` [ruby-core:91481] " merch-redmine
2019-02-08  0:38 ` [ruby-core:91485] " akr
2019-02-08  6:54 ` [ruby-core:91489] [Ruby trunk Feature#15592] mode where "autoload" behaves like an immediate "require" fxn
2019-02-09 13:36 ` akr [this message]
2019-02-09 17:47 ` [ruby-core:91502] " fxn
2019-02-10 15:20 ` [ruby-core:91504] " Greg.mpls

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-list from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

  List information: https://www.ruby-lang.org/en/community/mailing-lists/

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=redmine.journal-76761.20190209133635.04a1b02e829ca335@ruby-lang.org \
    --to=ruby-core@ruby-lang.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).