From: Bryan Donlan <bdonlan@gmail.com>
Date: Fri, 15 Aug 2008 19:31:25 GMT
Subject: Re: [PATCH] Git.pm: Make File::Spec and File::Temp requirement lazy
Message-ID: <3e8340490808151231p700ca76fub16700708f2942bb@mail.gmail.com>
In-Reply-To: <1218813032-18203-1-git-send-email-marcus@griep.us>

On Fri, Aug 15, 2008 at 11:10 AM, Marcus Griep <marcus@griep.us> wrote:
> This will ensure that the API at large is accessible to nearly
> all Perl versions, while only the temp file caching API is tied to
> the File::Temp and File::Spec modules being available.
>
> Signed-off-by: Marcus Griep <marcus@griep.us>
> ---
>
>  Eric Wong wrote:
>  > What about just lazy requiring inside _temp_cache() so it
>  > won't get loaded by folks that don't need it? (completely untested):
>  >
>  >              eval { require File::Temp };
>  >              if ($@) {
>  >                      throw Error::Simple("couldn't require File::Temp: $@");
>  >              }
>  >              eval { require File::Spec };
>  >              if ($@) {
>  >                      throw Error::Simple("couldn't require File::Spec: $@");
>  >              }
>  >
>  > It'll also remove the minor performance hit CGI/gitweb users got since
>  > we won't load these extra modules during startup.
>
>  This recommendation is implemented with this patch, but in such a way that only
>  the first test will be used, and that result cached.  That way we aren't doing
>  a compile _every_ time we want a temporary file, just the first time.

perl's 'require' will only attempt to load a module the first time it
is used; subsequent attempts will result in a quick no-op, so there's
no need to further cache. Unless you mean to cache a negative result?
In which case, the require should be quite fast, as it'll quickly get
a 'file not found' error without needing to do any compilation.