[bug#39384] [PATCH] gnu: Add emacs-rg.

[Efraim Flashner <efraim@flashner.co.il>]

[-- Attachment #1: Type: text/plain, Size: 2461 bytes --]

On Sat, Feb 08, 2020 at 04:23:08PM -0600, LaFreniere, Joseph wrote:
> 
> Efraim Flashner <efraim@flashner.co.il> writes:
> > On Thu, Feb 06, 2020 at 06:47:23PM -0600, LaFreniere, Joseph wrote:
> > > 
> > > Marius Bakke <mbakke@fastmail.com> writes:
> > > > "LaFreniere\, Joseph" <joseph@lafreniere.xyz> writes:
> > > > > Ah, I see what you mean now.  But wouldn't hard-coding the > >
> > > path to
> > > > > ripgrep in that way prevent the package from being able to > >
> > > use
> > > > > remote systems' ripgrep binaries when running over TRAMP?
> > > >
> > > > Perhaps we could patch [emacs-rg] to do both?  Use the store >
> > > prefix if
> > > > it
> > > > exists, and fall back to searching in PATH?
> > > 
> > > What would be the advantage of that over just searching PATH to
> > > start with?
> > 
> > It will still work even if you don't have ripgrep specifically
> > installed.
> 
> Can you point me to the Guix documentation where the functionality you're
> describing is explained?  I have read through the description of package
> inputs in section 6.2.1 of Guix's manual, but I still do not explain what
> advantage patching the search path offers.

I'm not sure I can find a spot in the manual where it is detailed. It
comes down to the difference between "search for this program in PATH"
and "call this program located at this location". By calling the rg
at it's exact path rg doesn't need to be installed directly.

> My understanding is that if we want to preserve both local and
> remote-via-TRAMP functionality, we can either
> - just include ripgrep as a propagated input, or
> - include ripgrep as a propagated input _and_ patch the package to  look for
> ripgrep in a hardcoded location (for local) as well as  PATH (for TRAMP).

The second option is to include ripgrep as an input and patch the
package to look for it at a hardcoded location (for local) as well as
PATH (for TRAMP).

> Both options would have ripgrep included as propagated input.  As soon as
> ripgrep is installed in a user's profile, its binary will be available on
> PATH.  If that is correct, then I don't see any advantage to patching in a
> hardcoded path to ripgrep.
> 
> --
> Joseph LaFreniere

-- 
Efraim Flashner   <efraim@flashner.co.il>   אפרים פלשנר
GPG key = A28B F40C 3E55 1372 662D  14F7 41AA E7DC CA3D 8351
Confidentiality cannot be guaranteed on emails sent or received unencrypted

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]