package not needed to byte-compile

package not needed to byte-compile

[Joe Riel]

Is there a way to tell Emacs that package B
is required to run package A, but not to 
compile package A?   Using (require B)
causes an error during byte-compilation because the
package is not necessarily available then.


-- 
Joe Riel

Re: package not needed to byte-compile

[Joe Riel]

On Sun, 22 Feb 2015 19:07:02 -0800
Joe Riel <joer@san.rr.com> wrote:

> Is there a way to tell Emacs that package B
> is required to run package A, but not to 
> compile package A?   Using (require B)
> causes an error during byte-compilation because the
> package is not necessarily available then.

Apologies for the duplication; I thought my first post
would be rejected because it came from an account
not registered with this group.


-- 
Joe Riel

Re: package not needed to byte-compile

[Michael Heerdegen]

Joe Riel <joer@san.rr.com> writes:

> Is there a way to tell Emacs that package B is required to run package
> A, but not to compile package A?  Using (require B) causes an error
> during byte-compilation because the package is not necessarily
> available then.

I guess something like this, for example:

--8<---------------cut here---------------start------------->8---
(defun a-require-b ()
  (require 'b))

(a-require-b)
--8<---------------cut here---------------end--------------->8---

You can put a require call into any defun if you like (entry points to
code that makes use of B), that is not uncommon.  Of course A should not
make use of any macros defined in B.

If the whole B makes use of A's functions however, a toplevel require is
better.

If you avoid a toplevel require, you can use declare-function to get rid
of the compiler warnings about undefined functions.

Michael.

package not needed to byte-compile

[Joe Riel]

Is there a way to tell Emacs that package B
is required to run package A, but not to 
compile package A?   Using

(require B)

causes an error during byte-compilation because the
package is not available then.

-- 
Joe Riel

RE: package not needed to byte-compile

[Drew Adams]

> Is there a way to tell Emacs that package B
> is required to run package A, but not to
> compile package A?   Using (require B)
> causes an error during byte-compilation because the
> package is not available then.

It's not clear (to me) what you're asking.  `require' does not
byte-compile anything.  If A needs B then putting (require 'B)
in A loads B.  That compiles neither A nor B.

If you don't want to compile A, don't compile it.  Likewise, B.

Causes an error during byte compilation of what?  You said
that you do not want to compile A.  What error?

Your question is completely unclear to me.  I suggest that
you start over.  Tell us (a) what you did, specifically,
(b) what happened, specifically, including what error you
saw, and (c) what you expected to happen instead.

Re: package not needed to byte-compile

[Stefan Monnier]

> Using
> (require B)
> causes an error during byte-compilation because the
> package is not available then.

I've used

   (if t (require B))           ;Don't load during compilation.

when I needed that.


        Stefan

Re: package not needed to byte-compile

[Michael Heerdegen]

Stefan Monnier <monnier@iro.umontreal.ca> writes:

>    (if t (require B))           ;Don't load during compilation.

The rule is that only top-level `require' occurrences are special and
cause files to be loaded when compiling, right?  Because, I can't find
that fact spelled out in the docs.

Re: package not needed to byte-compile

[Stefan Monnier]

>> (if t (require B))           ;Don't load during compilation.
> The rule is that only top-level `require' occurrences are special and
> cause files to be loaded when compiling, right?

That's right.  Tho, "within a progn" or "within a macro which expands to
a progn" still counts as top-level.


        Stefan

Re: package not needed to byte-compile

[Michael Heerdegen]

Stefan Monnier <monnier@iro.umontreal.ca> writes:

> >> (if t (require B))           ;Don't load during compilation.
> > The rule is that only top-level `require' occurrences are special and
> > cause files to be loaded when compiling, right?
>
> That's right.  Tho, "within a progn" or "within a macro which expands to
> a progn" still counts as top-level.

Good to know, thanks.

I now also found out where it is described in the docs: it's discussed
in (info "(elisp) Named Features").