gptel
gptel copied to clipboard
gptel-org-branching-context requires Org 9.7 (unreleased)
Hi Karthik, first, thanks for gptel. It is truly wonderful.
It seems that gptel-org-branching-context
depends on the function org-element-lineage-map
, which is not included in the current release of Org (9.6.27), at least as far as I can tell.
It seems that the file containing that function, org-element-ast.el, is in org's main branch, but 9.6 is not built with it. I do not know why it is excluded, at least in my case.
Might the branching-context feature be adapted (without too much trouble) for those of us on just this side of org's bleeding edge? Thanks for considering it, in any case.
Flying by the seat of my pants here -- I bundled org-element-lineage-map
with gptel, to be evaluated at compile time if it's not natively available. Please let me know if it works, or if there are transitive dependencies that are also missing.
Looks like there are quite a few other functions that would also need to be bundled:
org-element-parent
org-element-begin
org-element-type
org-element-type-p
org-element-secondary-p
Not sure if bundling all these is something you would want to do.
In the meantime, I've just defined all these in my init file and gptel-org-branching-context
works great.
karthink @.***> writes:
Flying by the seat of my pants here -- I bundled org-element-lineage-map with gptel, to be evaluated at compile time if it's not natively available. Please let me know if it works, or if there are transitive dependencies that are also missing.
— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.
Oof, my fix was shortsighted then, as I expected.
I'm not sure what I can do here. I can throw a warning when setting gptel-org-branching-context
if the org version is not high enough, and disable it. But this warning will only fire if setting the variable using setopt or the customization menu, which I doubt most users are using.
Or maybe instead of a warning when setting the variable, a warning when calling gptel-send in a context when a user might expect it to work, but they don’t have the right org version. Or even simpler (and what I would do) just a note about the org version requirement in the doc string. Dont think it’s worth much more effort than that, honestly.
From: karthink @.>Sent: Sunday, April 28, 2024 06:22To: karthink/gptel @.>Cc: Grant Rosson @.>; Author @.>Subject: Re: [karthink/gptel] gptel-org-branching-context requires Org 9.7 (unreleased) (Issue #294) Oof, my fix was shortsighted then, as I expected. I'm not sure what I can do here. I can throw a warning when setting gptel-org-branching-context if the org version is not high enough, and disable it. But this warning will only fire if setting the variable using setopt or the customization menu, which I doubt most users are using.
—Reply to this email directly, view it on GitHub, or unsubscribe.You are receiving this because you authored the thread.Message ID: @.***>
Perhaps someone would find it useful, if only temporarily, if you mentioned this package, which provides a similar feature (as a light wrapper around gptel) using released versions of org.
Thank you @ultronozm.
@localauthor I added a warning when gptel-org-branching-context
is set if it can't find org-element-lineage-map
. I don't check for the availability of org-element-parent
and the rest, but I think anyone updating from an older Org version for this feature will update to the tip of the Org main branch so this should suffice.
I'm running emacs.29.4
+ latest spacemacs develop
+ org.9.7.8
+ gptel-20240718.2048
and at doing an actual request send to openai
server I'm getting an error:
gptel-org--create-prompt: Symbol's function definition is void: gptel-org--element-lineage-map
The problem gets fixed after enclosing eval-when-compile
is commented out
I experience the problem since very beginning of my journey with gptel
, ie since emacs 29.3
org.9.6.15
and gptel
from about 2 months ago
should I open new issue?
@vibrys Thanks for letting me know. I've tried to fix it in the latest commit.
@karthink, thank you for prompt response. Although your newest fix hasn't been propagated yet, I've taken it manually to my copy of a gptel-org package to see if it has chance to help and it looks like the fix is doing the job.
thank you! also thank you for gptel!