| Age | Commit message (Collapse) | Author | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
Since this job is done in the process sentinel, the clean up is also
triggered when the Scheme process exits unexpectedly, deleting any
traces the dying guy might have left.  I added a flag to control the
behaviour, but upon reflection the old behaviour seems wrong and i've
defaulted to the new one.  This one should fix #251.
 | 
 | 
 | 
 | 
Generated by simply copying README.org.  The trick of making README a
symbolic link was, well, a trick, and seems to confuse poor Gitlab.
 | 
 | 
Probably not a totally smart move, since sooner or later gitlab's
going to fall as github did (for a fruity company would be my bet).
But oh well, at least we can export a CSV of the issues!
 | 
 | 
And we also take the chance to let add-to-list do its job of not
adding duplicates.
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
and make github happier in the process (cf. github issue #243)
 | 
 | 
And thus avoiding having to generate markdown (cf. github issue #243)
 | 
 | 
 | 
 | 
Okay, i must confess it's sometimes handy to restart the REPL before
compiling a file (the proverbial clean slate and all).  And we already
have geiser-restart-repl, so combining the two things when C-u happens
was not really difficult.
 | 
 | 
Savannah downloads is often broken, and the news page doesn't really
add any value.
 | 
 | 
 | 
 | 
 | 
 | 
Looks like the arity of that function changed at some point between 24
and 25.  It also looks like people still use emacs 24 (see issue #236),
so here we go.
 | 
 | 
 | 
 | 
Those two procedures are gone in these 2.2 times, and things like
autodoc and xref were broken as a result.  With Andy's help,
apparently good enough approximations of their functionality are now
in place: let's see how they go.
 | 
 | 
When constructing the completion table for minibuffer prompts via
`completion-table-dynamic', we were forgetting to tell emacs to
perform the completion lookup with the original (scheme) buffer as its
current buffer.  As a result, the actual completion function wasn't
able to find the REPL connection and everything when down in flames
with an exception.
 | 
 | 
 | 
 | 
We can probably do away with this entirely; but for now, make it a custom.
 | 
 | 
It now _slows_ performance rather than improves it. Removing it speeds
up the issue described in jaor/geiser#174
 | 
 | 
Necessary for jaor/geiser#174
 | 
 | 
Removed all of the symbol-interning code, and in the process greatly
reduced the amount of CPU time.
Should resolve jaor/geiser#174
 | 
 | 
It's only using string-empty-p from there, which is a function whose
name is almost as long as its implementation, so we better try to be
compatible with older emacsen (subr-x was introduced in emacs 24.4).
 | 
 | 
 | 
 | 
For some reason, one of our users is experiencing point jumps when
calling `geiser-set-scheme'.  A save-excursion is all that's needed,
even though it *shouldn't* be needed in the first place.
 | 
 | 
`install-package` -> `package-install`
 | 
 | 
Chicken Scheme's apropos egg changed the identifiers in the return value
of the `apropos-information-list` from `module#name` to `(module
. name)`. This commit adds support for the new identifiers.
 | 
 | 
* elisp/geiser-guile.el (guile--manual-look-up): Change parameters for
  info-lookup-symbol to string and 'scheme-mode.  Fixes lookup.
 | 
 | 
 | 
 | 
as per the discussion at: https://github.com/jaor/geiser/issues/183
 | 
 | 
 | 
 | 
 | 
 | 
 | 
 | 
geiser-mode-eval-to-buffer-transformer will take 2 argments:
errstring and result
when eval-to-buffer, the result will be transformed by this procedure
e.g.
(setq geiser-mode-eval-to-buffer-transformer
      (lambda (estring x)
	(let ((l (length x))
	      (p (seq-position x ?\n)))
	  (if (and p (< (+ 1 p) l))
	      (format "\n#| %s%s\n  |#" estring x)
	    (format ";;=> %s%s" estring x)))))
 | 
 | 
 |