| Age | Commit message (Collapse) | Author |
|---|
|
We were using a comint-get-old-input function that was including the
prompt in its returned value. This was no problem most of the time
because we don't use comint-send-input before the process mark, but
there's another circumstance under which comint-get-old-input is
called, namely, when reaching the end of the input history. When
history is exhausted, the "old input" is inserted (go figure), and we
were inserting a prompt, wreaking havoc with its read-only-ness.
|
|
My previous attempt was bogus! Also, i'm taking advantage of our
sending queries serially to simplify transaction queue processing, and
to clean after ourserlves on error.
|
|
Let's use the scheme reader instead of half-assed regular expressions
and special-casing.
|
|
|
|
Apparently, quack makes (thing-at-point 'symbol) to return "#" when
point is looking at a hash symbol, causing havoc to geiser. This patch
defends us about this behaviour.
|
|
|
|
|
|
We cannot consistently maintain a local cache, because of
re-evaluations of external symbols will go unnoticed. The new strategy
(remembering only the latest signatures) mostly works, although it
introduces a bit of extra flickering every now and then.
A global cache is perhaps worth considering.
|
|
Only display an activation message when called interactively.
|
|
The message and the evaluated expression where getting mixed for Guile.
|
|
|
|
A piece of eldoc advice shamelessly lifted from slime.
|
|
|
|
|
|
Namely, those with non-letters in their name, because we were using
"%S" instead of "%s" to stringify uninterned symbols.
|
|
I've observed that autodoc requests in connections to tekuti processes
often miss their deadline. This was causing serialization problems,
fixed (i think) by this patch; but we still have the problem of too
many misses, which could be fixed by making autodoc asynchronous (and
we probably need this for really remote connections anyway).
|
|
Maybe i should bite the bullet and use ##autoload cookies, but then
geiser.el would need to be a generated file, breaking
out-of-the-boxness (unless i update it by hand). Sticking to manual
maintenance for now.
|
|
If quack-mode is active, we leave additional font locking to it.
|
|
A couple functions to manage logs, and a fix to the definition of
geiser-messages-mode: do not ever kill all local variables in a
derived mode definition!
|
|
I've eliminated the annoying *spurious* popup produced by tq, so that
Geiser can recover from error conditions arising from unexpected
input to a transaction queue that has no active transaction. We now
log the offending input and keep going.
Greg, over at racket's list, reported such a happening when leaving
the REPL after C-u C-c C-z; but i've been unable to reproduce the
problem. Probably, we have a bug lurking somewhere that this patch
doesn't fix, but at least it should work as a palliative.
|
|
This is Emacs, after all.
|
|
You won't believe it, but people write procedures spanning several
pages, even in Scheme. Where's the art and the poetry of programming?
|
|
... and i haven't yet refactored `geiser-syntax--scan-locals', oh my.
|
|
... using the new non-interning reader. Plus scanning for case-lambda
and syntax-rules. `geiser-syntax--scan-locals' is in danger of
refactoring, specially if we add support for let-values.
|
|
It was relying on symbol equality, and not dealing correctly with
keyword arguments in this new external world. In the process, some
cleanups to the scheme data display code.
|
|
We were still internalizing scheme _keywords_ in the elisp reader.
|
|
|
|
Since, you know, module names are now uninterned symbols.
|
|
We avoid using elisp's read for symbols, reading uninterned ones
instead. And then, we cannot use symbols as keys in responses from
scheme: we're using strings instead.
|
|
These ones seem safe: the resulting symbol is not compared for
equality anywhere.
|
|
We avoid calling symbol-at-point, and keep the cached signatures with
strings as keys.
|
|
... which interns the symbol in the global obarray: rather unfriendly.
We still need to remove a few calls to that beast, and avoid intern in
the scheme reader.
|
|
We were adding extra spaces to function signatures.
|
|
|
|
This reverts commit 801422d1558f488059ede4f9abab5163ca610900.
We cannot blindly substitute make-symbol for intern in the scheme
reader, because we rely on symbol equality elsewhere, often. The fix
will have to be much more careful.
|
|
We were calling `intern' instead of `make-symbol', polluting emacs'
obarray.
|
|
When the symbol is imported and re-exported by a second module, we
display its definition name and original module, besides the name of
the module re-exporting it.
|
|
|
|
But i should really refactor this: module and value are (or can be)
already available in the response coming from Scheme.
|
|
Just justifying and indenting them.
|
|
... and actually using it to implement geiser-smart-tab-mode. Always
nice to un-reinvent-the-wheel.
|
|
Besides removing code i didn't understand that well, we bring in
goodies such as partial completion. Jolly.
|
|
Was a real bug, after all, and quite reproducible. Sending an ,use
metacommand was not returning a prompt, and we were waiting for ever
to start (or almost). Now, connect-to-guile is not only right, but
spiffy again.
|
|
|
|
|
|
|
|
We have a new "manual lookup" command, and Racket now displays a doc
browser buffer for help with a button activating it. In the process,
we've cleaned-up a little mess in geiser-eval.el and geiser-doc.el,
and refactored the affected Racket modules.
Next in line is providing manual lookup for Guile.
|
|
|
|
geiser-repl was missing a (require 'geiser-doc) that was making things
go pretty awry for compiled geiser on os x (emacs 23.2.20), but
nowhere else, for reasons that escape me.
Issue was, the popup buffer macros were not seen. Go figure.
|
|
Details, details.
|
