bug#2061: Fwd: Loading .el when newer than .elc

bug#2061: Fwd: Loading .el when newer than .elc

[Reuben Thomas]

[-- Attachment #1: Type: text/plain, Size: 2341 bytes --]

I just got annoyed enough about missing a warning that an .el file is newer
than an .elc file to want to file a bug report, and sure enough found this
issue has already been discussed.

To summarize the discussion so far, the following reasons have been
advanced for not wanting automatically to prefer a newer .el over an older
.elc:

1. One might be working on the .el file, and not want to load it yet.

2. This is a bug, which should be fixed by recompiling the .el.

Several workarounds have been proposed for those who want to use the newest
version of the .el; they all involve recompiling the file (which some
people complained slowed things down too much, and others pointed out could
lead to loops when working on a large project), and all involve taking
action for specific files, e.g. per-project, or per-file, or per-directory.

I think it's worth looking at those objections again in a modern context:

1. We don't work on deployed code any more (or shouldn't!). We use version
control systems, we run tests before deploying. When I fiddle with a file,
I always want my changes to take effect instantly: either I'm working on
code I'm developing, or I'm trying a fix on a deployed file (and am willing
to keep the bits if it breaks!).

2. GNU distributions go to some lengths to ensure that all system-installed
.el files are compiled and up-to-date. el-get does the same for
user-installed packages. There's little else.

Meanwhile, the workarounds require almost as much effort to install and
maintain as simply remembering to byte-compile everything.

Personally, I have wasted hours by thinking I'm loading a new .el when I'm
loading an old .elc, plus going through various recompilation dances when I
do remember. The time that defaulting to .elc has saved me is minimal,
owing to Debian+el-get taking care of recompilation.

Hence, could we have a preference that reverses the behavior, i.e. when the
.el is newer, it is loaded, and a warning emitted? The default behavior
should of course remain the same. It's a simple, uniform, discoverable
solution. I'd happily work up a patch myself. The preference could be used
in .dir-locals.el by users who preferred to switch it on only for, say,
per-user site-lisp directories or project directories.

How about: load-el-when-newer (defaults to nil) ?

--
http://rrt.sc3d.org

[-- Attachment #2: Type: text/html, Size: 2528 bytes --]

bug#2061: Fwd: Loading .el when newer than .elc

[Eli Zaretskii]

> Date: Tue, 11 Feb 2014 12:48:42 +0000
> From: Reuben Thomas <rrt@sc3d.org>
> 
> How about: load-el-when-newer (defaults to nil) ?

From etc/NEWS:

  ** New option `load-prefer-newer' affects how the `load' function chooses
  the file to load.  If this is non-nil, then when both .el and .elc
  versions of a file exist, and the caller did not explicitly specify
  which one to load, then the newer file is loaded.  The default, nil,
  means to always load the .elc file.

bug#2061: Fwd: Loading .el when newer than .elc

[Reuben Thomas]

[-- Attachment #1: Type: text/plain, Size: 738 bytes --]

On 11 February 2014 16:14, Eli Zaretskii <eliz@gnu.org> wrote:

> > Date: Tue, 11 Feb 2014 12:48:42 +0000
> > From: Reuben Thomas <rrt@sc3d.org>
> >
> > How about: load-el-when-newer (defaults to nil) ?
>
> From etc/NEWS:
>
>   ** New option `load-prefer-newer' affects how the `load' function chooses
>   the file to load.  If this is non-nil, then when both .el and .elc
>   versions of a file exist, and the caller did not explicitly specify
>   which one to load, then the newer file is loaded.  The default, nil,
>   means to always load the .elc file.
>

Joy! I'm sorry my timing was bad, in that I looked into this after it had
been committed, but before it had been released. Thanks for the
intelligence.

-- 
http://rrt.sc3d.org

[-- Attachment #2: Type: text/html, Size: 1244 bytes --]

bug#2061: Fwd: Loading .el when newer than .elc

[Glenn Morris]

Reuben Thomas wrote:

> Joy! I'm sorry my timing was bad, in that I looked into this after it had
> been committed, but before it had been released.

That's fine, but why not read the report, especially the close message,
before reopening it?

http://debbugs.gnu.org/cgi/bugreport.cgi?bug=2061#106

seems clear to me.

Now the OP will wonder why his bug was opened and closed again.

bug#2061: Fwd: Loading .el when newer than .elc

[Reuben Thomas]

[-- Attachment #1: Type: text/plain, Size: 608 bytes --]

On 11 February 2014 17:03, Glenn Morris <rgm@gnu.org> wrote:

> Reuben Thomas wrote:
>
> > Joy! I'm sorry my timing was bad, in that I looked into this after it had
> > been committed, but before it had been released.
>
> That's fine, but why not read the report, especially the close message,
> before reopening it?
>
> http://debbugs.gnu.org/cgi/bugreport.cgi?bug=2061#106
>
> seems clear to me.
>

You're right, it is. However, I read the bug report on the mailing list
archive, and didn't find the close message there. The moral is to use the
debbugs interface for completeness.

-- 
http://rrt.sc3d.org

[-- Attachment #2: Type: text/html, Size: 1120 bytes --]