[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: What's the right way to define a custom info path.
From: |
Pierre Téchoueyres |
Subject: |
Re: What's the right way to define a custom info path. |
Date: |
Wed, 25 Nov 2015 23:13:10 +0100 |
User-agent: |
KNode/4.14.10 |
Eli Zaretskii wrote:
>> From: Pierre Téchoueyres <pierre.techoueyres@free.fr>
>> Date: Wed, 25 Nov 2015 20:54:25 +0100
>>
>> > No, there's no bug here, AFAICT. If you want to set up
>> > Info-directory-list or Info-additional-directory-list, you must load
>> > info.elc first. (But again, I don't recommend going that way.)
>>
>> But these two variables could be modified with the custom machinery, and
>> so without requiring info[.elc] aren't they ?
>
> Yes, you could do that. But I interpreted your message as a request
> to set them up in Lisp, not via Custom.
>
I've tried both aproaches. First with customize, but I found that package
initialization discarded what I had set.
I expected that package init and customize have well worked together.
>> But I understand your advice that doing that is discouraged.
>
> No, it's not discouraged. It just is harder to set up correctly,
> whereas the semantics of INFOPATH is simple.
I'll try this tomorrow. But I guess I should set all info dir, even the ones
installed by packages. I must check that.
>
>> Second, the default value for Info-default-directory-list (as computed by
>> the defcustom in info.el) is ("%emacs_dir/info") on my windows install.
>> Is this the expected behaviour ?
>
> Yes. That value is never used.
>
>> this value is obviously overridden by info-
>> initialize and become, in my install, ("c:/programmes/emacs/info").
>> Again is this the expected behaviour ?
>
> Yes.
>
>> Third, in the windows patform (substitute-env-vars "%emacs_dir%") doesn't
>> produce "c:/programmes/emacs" as I expected. But (substitute-env-vars
>> "$emacs_dir") do the expansion. Is this the expected behaviour ?
>
> Yes. substitute-env-vars supports the Unix style of environment
> variables.
>
>> Is it that we should not offer a version that performs primary
>> processing ?
>
> We could, but why bother? When info loads, it recomputes the value
> according to where Emacs was installed.
To offer a substitute-env-vars with windows style ?