* lisp/erc/erc-backend.el (erc-server-connected): Revise doc string.
(erc-server-reconnect, erc-server-JOIN): Reuse original ID param from
the first connection when calling `erc-open'.
(erc-server-NICK): Apply same name generation process used by
`erc-open'; except here, do so for the purpose of "re-nicking".
Update network identifier and maybe buffer names after a user's own
nick changes.
* lisp/erc/erc-networks.el (erc-networks--id, erc-networks--id-fixed,
erc-networks--id-qualifying): Define new set of structs to contain all
info relevant to specifying a unique identifier for a network context.
Add a new variable `erc-networks--id' to store a local reference to a
`erc-networks--id' object, shared among all buffers in a logical
session.
(erc-networks--id-given, erc-networks--id-create,
erc-networks--id-on-connect, erc-networks--id--equal-p,
erc-networks--id-qualifying-init-parts,
erc-networks--id-qualifying-init-symbol,
erc-networks--id-qualifying-grow-id,
erc-networks--id-qualifying-reset-id,
erc-networks--id-qualifying-prefix-length,
erc-networks--id-qualifying-update, erc-networks--id-reload,
erc-networks--id-ensure-comparable, erc-networks--id-sort-buffers):
Add new functions to support management of `erc-networks--id' struct
instances.
(erc-networks--id-sep): New variable for to help when formatting
buffer names.
(erc-obsolete-var): Define new generic context rewriter.
(erc-networks-shrink-ids-and-buffer-names,
erc-networks--refresh-buffer-names,
erc-networks--shrink-ids-and-buffer-names-any): Add functions to
reassess all network IDs and shrink them if necessary along with
affected buffer names. Also add function to rename buffers so that
their names are unique. Register these on all three of ERC's
kill-buffer hooks because an orphaned target buffer is enough to keep
its session alive.
(erc-networks-rename-surviving-target-buffer): Add new function that
renames a target buffer when it becomes the sole bearer of a name
based on a target that has become unique across all sessions and, in
most cases, all networks. IOW, remove the @NETWORK-ID suffix from the
last remaining channel or query buffer after its namesakes have all
been killed off. Register this function with ERC's target-related
kill-buffer hooks.
(erc-networks--examine-targets): Add new utility function that visits
all ERC buffers and runs callbacks when a buffer-name collision is
encountered.
(erc-networks--qualified-sep): Add constant to hold separator between
target and suffix.
(erc-networks--construct-target-buffer-name,
erc-networks--ensure-unique-target-buffer-name,
erc-networks--ensure-unique-server-buffer-name,
erc-networks--maybe-update-buffer-name): Add helpers to support
`erc-networks--reconcile-buffer-names' and friends.
(erc-networks--reconcile-buffer-names): Add new buffer-naming strategy
function and helper for `erc-generate-new-buffer-name' that only run
in target buffers.
(erc-determine-network, erc-networks--determine): Deprecate former and
partially replace with latter, which demotes RPL_ISUPPORT-derived
NETWORK name to fallback in favor of known `erc-networks-alist'
members as part of shift to network-based connection-identity policy.
Return sentinel on failure. Expect `erc-server-announced-name' to be
set, and signal when it's not.
(erc-networks--name-missing-sentinel): Value returned when new
function `erc-networks--determine' fails to find network name. The
rationale for not making this customizable is that the value signifies
the pathological case where a user of an uncommon IRC setup has not
yet set a mapping from announced- to network name. And the chances of
there being multiple unknown networks is low.
(erc-set-network-name, erc-networks--set-name): Deprecate former and
partially replace with latter. Ding with helpful message, and don't
set `erc-network' when network name is not found.
(erc-networks--ensure-announced): Add new fallback function to ensure
`erc-server-announced-name' is set. Register with post-MOTD hooks.
(erc-unset-network-name): Deprecate function unused internally.
(erc-networks--insert-transplanted-content,
erc-networks--reclaim-orphaned-target-buffers,
erc-networks--copy-over-server-buffer-contents,
erc--update-server-identity): Add helpers for
`erc-networks--rename-server-buffer'. The first re-associates all
existing target buffers that ought to be owned by the new server
process. The second grabs buffer text from an old, dead server buffer
before killing it. It then inserts that text above everything in the
current, replacement server buffer. The other two massage the IDs of
related sessions, possibly renaming them as well. They may also
uniquify the current session's network ID.
(erc-networks--init-identity): Add new function to perform one-time
session-related setup. This could be combined with
`erc-set-network-name'.
(erc-networks--rename-server-buffer): Add new function to replace
`erc-unset-network-name' as default `erc-disconnected-hook' member;
renames server buffers once network is discovered; added to/removed
from `erc-after-connect' hook on `erc-networks' minor mode.
(erc-networks--bouncer-targets): Add constant to hold target symbols
of well known bouncer-configuration bots.
(erc-networks-on-MOTD-end): Add primary network-context handler to run
on 376/422 functions, just before logical connection is officially
established.
(erc-networks-enable, erc-networks-mode): Register main network-setup
handler with 376/422 hooks.
* lisp/erc/erc.el (erc-rename-buffers): Change this option's default
to t, remove the only instance where it's actually used, and make it
an obsolete variable.
(erc-reuse-buffers): Make this an obsolete variable, but take pains to
ensure its pre-28.1 behavior is preserved. That is, undo the
regression involving unwanted automatic reassociation of channel
buffers during joins, which arrived in ERC 5.4 and effectively
inverted the meaning of this variable, when nil, for channel buffers,
all without accompanying documentation or announcement.
(erc-generate-new-buffer-name): Replace current policy of appending a
slash and the invocation host name. Favor instead temporary names for
server buffers and network-based uniquifying suffixes for channels and
query buffers. Fall back to the TCP host:port<n> convention when
necessary. Accept additional optional params after the others.
(erc-get-buffer-create): Don't generate a new name when reconnecting,
just return the same buffer. `erc-open' starts from a clean slate
anyway, so this just keeps things simple. Also add optional params.
(erc-open): Add new ID param to for a network identifier explicitly
passed to an entry-point command. This is stored in the `given' slot
of the `erc-network--id' object. Also initialize the latter in new
connections and otherwise copy it over. As part of the push to recast
erc-networks.el as an essential library, set `erc-network' explicitly,
when known, rather than via hooks.
(erc, erc-tls): Add new ID keyword parameter and pass it to
`erc-open'.
(erc-log-irc-protocol): Use `erc--network-id' instead of the function
`erc-network' to determine preferred peer name.
(erc-format-target-and/or-network): This is called frequently from
mode-line updates, so renaming buffers here is not ideal. Instead, do
so in `erc-networks--rename-server-buffer'.
(erc-kill-server-hook): Add `erc-networks-shrink-ids-and-buffer-names'
as default member.
(erc-kill-channel-hook, erc-kill-buffer-hook): Add
`erc-networks-shrink-ids-and-buffer-names' and
`erc-networks-rename-surviving-target-buffer' as default member.
* test/lisp/erc/erc-tests.el (erc-log-irc-protocol): Use network-ID
focused internal API.
* test/lisp/erc/erc-networks-tests.el: Add new file that includes
tests for the above network-ID focused functions.
See bug#48598 for background on all of the above.
* test/lisp/erc/resources/erc-d/erc-d.el: New file.
* test/lisp/erc/resources/erc-d/erc-d-u.el: New file.
* test/lisp/erc/resources/erc-d/erc-d-i.el: New file.
* test/lisp/erc/resources/erc-d/erc-d-t.el: New file.
* test/lisp/erc/resources/erc-d/erc-d-tests.el: New file.
* test/lisp/erc/erc-scenarios-internal.el: New file to serve as
discoverable proxy for erc-d-tests.
* lisp/erc/erc.el (erc--target, erc--target-channel,
erc--target-channel-local): Add new structs to hold info on a buffer's
target; stored in a local variable of the same name.
(erc--target-from-string): Add standalone constructor for
`erc--target'.
(erc--default-target): Add temporary internal getter to ease
transition to `erc--target' everywhere.
(erc-open): Create above items in non-server buffers.
* lisp/erc/erc-backend.el (erc-server-NICK): Recreate `erc--target'
when necessary.
* lisp/erc/erc.el (erc--valid-local-channel-p): Add internal helper to
determine whether an IRC channel is local according to its network's
CHANTYPES ISUPPORT parameter.
* test/lisp/erc/erc-tests.el (erc--valid-local-channel-p): Add test
for this helper.
* lisp/erc/erc.el (erc-downcase, erc--casemapping-rfc1459-strict,
erc--casemapping-rfc1459): Add new translation tables for the latter
two mappings and use them in `erc-downcase'.
* test/lisp/erc/erc-tests.el: Add test for `erc-downcase'.
* lisp/erc/erc-backend (erc--isupport-params): Add new variable to
hold a hashmap of parsed `erc-server-parameters' in a more useful
format. But keep `erc-server-parameters' around for public use. We
currently lack dedicated local variables for certain discovered IRC
session properties, such as what prefix characters are supported for
channels, etc. And the truth of this needs querying many times per
second at various points. As such, caching here seems justified but
can be easily removed if deemed otherwise because all ingredients are
internal.
(erc--parse-isupport-value): Add helper function that parses an
ISUPPORT value and returns the component parts with backslash-x hex
escapes removed. This can probably use some streamlining.
(erc--with-memoization): Add compat alias for use in internal ISUPPORT
getter. Should be moved to `erc-compat.el' when that library is fully
reincorporated.
(erc--get-isupport-entry): Add internal getter to look up ISUPPORT
items.
(erc-server-005): Treat `erc-server-response' "command args" field as
read-only. Previously, this field was set to nil after processing,
which was unhelpful to other parts of the library. Also call above
mentioned helper to parse values. And add some bookkeeping to handle
negation.
* lisp/erc/erc-capab.el (erc-capab-identify-send-messages): Use
internal ISUPPORT getter.
* lisp/erc/erc.el (erc-cmd-NICK, erc-parse-prefix,
erc-nickname-in-use): Use internal ISUPPORT getter.
* test/lisp/erc/erc-tests.el: Add tests for the above mentioned
changes in erc-backend.el.
* lisp/erc/erc-backend.el (erc--unhide-prompt, erc--hide-prompt,
erc--unhide-prompt-on-self-insert): Add functions to ensure prompt is
hidden on disconnect and shown when a user types /reconnect in a
disconnected server buffer.
(erc-process-sentinel): Register aforementioned function with
`pre-command-hook' when prompt is deleted after disconnecting.
(erc-server-PRIVMSG): Ensure prompt is showing when a new message
arrives from target.
* lisp/erc/erc.el (erc-hide-prompt): Repurpose unused option by
changing meaning slightly to mean "selectively hide prompt when
disconnected." Also delete obsolete, commented-out code that at some
point used this option in its prior incarnation.
(erc-prompt-hidden): Add new option to specify look of prompt when
hidden.
(erc-unhide-query-prompt): Add option to force-reveal query prompts on
reconnect.
(erc-open): Augment earlier reconnect-detection semantics by
incorporating `erc--server-reconnecting'. In existing buffers, remove
prompt-related hooks and reveal prompt, if necessary.
(erc-cmd-RECONNECT): Allow a user to reconnect when already
connected (by first disconnecting).
(erc-connection-established): Possibly unhide query prompts.
(Bug#54826)
* test/lisp/erc/erc-tests.el (erc-tests--test-prep,
erc-tests--set-fake-server-process): Factor out some common
buffer-prep boilerplate involving user input and the server process.
Shared with bug#54536.
* lisp/erc/erc.el: Declare needed variables exported by erc-backend.el
as special near the top of the file, and only require `erc-backend'
after providing `erc' as a feature at the very end.
* lisp/erc/erc-backend.el: Don't preemptively provide `erc-backend'.
* test/lisp/erc/erc-tests.el (erc--meta--backend-dependencies): Add
utility test to scrape for unused vars that may accumulate over time.
(Bug#54825)
* test/lisp/emacs-lisp/text-property-search-tests.el
(text-property-search-forward/point-at-beginning)
(text-property-search-backward/point-at-end): Rewrite to make tests
easier to read.
* lisp/x-dnd.el (x-dnd-handle-old-kde, x-dnd-handle-offix)
(x-dnd-handle-motif): Select window before handling drop, like
on Xdnd.
(x-dnd-convert-to-offix, x-dnd-do-offix-drop)
(x-dnd-handle-unsupported-drop): Accept local selection data and
use that instead.
* src/keyboard.c (kbd_buffer_get_event): Call unsupported drop
function with local selection data as 8th arg.
* src/xselect.c (x_get_local_selection): Accept new arg
`local_value'. All callers changed.
(Fx_get_local_selection): New function.
(syms_of_xselect): Update defsubrs.
* src/xterm.c (x_dnd_lose_ownership): New function.
(x_dnd_begin_drag_and_drop): Unless new variable is true, disown
XdndSelection after returning. This supposedly makes
drag-and-drop from guest to host work in VirtualBox without
causing pointer motion to become choppy afterwards.
(syms_of_xterm): New variable `x_dnd_preserve_selection_data'
and update doc string of `x-dnd-unsupported-drop-function'.
* test/lisp/dnd-tests.el (dnd-tests-begin-text-drag)
(dnd-tests-begin-file-drag, dnd-tests-begin-drag-files): Set new
variable to nil during tests.
* lisp/subr.el (plistp): New type predicate (bug#47427). This
referred to in the error message from plist-put: "Debugger
entered--Lisp error: (wrong-type-argument plistp (a b c))".
* lisp/eshell/esh-io.el (eshell-close-target): Send EOF 3 times.
* test/lisp/eshell/em-extpipe-tests.el (em-extpipe-tests--deftest):
Re-enable these tests on EMBA.
This patch is adapted by one from Ken Brown, who uncovered the reason
for this bug (bug#56025).
* lisp/eshell/em-glob.el (eshell-glob-convert): Return whether to
match directories only.
(eshell-glob-entries): Add ONLY-DIRS argument.
* test/lisp/eshell/em-glob-tests.el
(em-glob-test/match-any-directory): New test.
(em-glob-test/match-recursive)
(em-glob-test/match-recursive-follow-symlinks): Add test cases for
when "**/" or "***/" are the last components in a glob.
* etc/NEWS: Announce this change (bug#56227).
* lisp/eshell/em-glob.el (eshell-glob-recursive): New variable.
(eshell-glob-convert-1, eshell-glob-convert): New functions.
(eshell-extended-glob): Use 'eshell-glob-convert'.
(eshell-glob-entries): Adapt function to use pre-converted globs.
* test/lisp/eshell-em-glob-tests.el (em-glob-test/match-dot-files):
New test.
* lisp/eshell/em-pred.el (eshell-error-if-no-glob): Declare it.
(eshell-apply-modifiers): Add STRING-DESC argument and signal an error
if there are no matches and 'eshell-error-if-no-glob' is set.
(eshell-parse-arg-modifier): Pass modifier string to
'eshell-apply-modifiers'.
* test/lisp/eshell/em-pred-tests.el (eshell-eval-predicate): Simplify.
(em-pred-test/no-matches): New test.
* doc/misc/eshell.texi (Bugs and ideas): Remove todo entry about this
change.
It would be backwards-incompatible to error out on invalid literal key
substitutions. Consider this docstring fragment, where "\\`" should
have been escaped but wasn't:
"Example value: \"\\(?:\\`[#.]\\)\\|\\(?:[#~]\\'\\)\"."
If we error out, we can't display this docstring at all. However, it
is clearly better to display something in such cases, even if
suboptimal, than refusing to display anything at all.
* lisp/help.el (substitute-command-keys): Don't error out on invalid
literal key substitutions: just ignore them instead.
* test/lisp/help-tests.el
(help-tests-substitute-command-keys/literal-key-sequence-errors):
Delete test.
(help-tests-substitute-command-keys/literal-key-sequence-ignore-invalid):
New test.
* lisp/sqlite-mode.el (require):
* lisp/net/eudc.el (require):
* lisp/arc-mode.el (require): Require subr-x, since these files
are using macros from there.
* lisp/emacs-lisp/subr-x.el (with-memoization): Move from here...
* lisp/subr.el (with-memoization): ... to here, as it's used from
the preloaded cl-generic.el file.
* lisp/emacs-lisp/cl-generic.el (cl--generic-compiler): Don't use
the autoloaded `byte-compile' function during bootstrap.
(cl--generic-get-dispatcher): Don't require subr-x, either.
cl-generic has been preloaded since 2015, and most usages of it (in
preloaded files) work fine. In particular, using `cl-defgeneric' is
unproblematic. However, `cl-defmethod' would end up pulling in the
byte compiler (at load time), which would make it impossible to use
`cl-defmethod' in pre-loaded files, and this change fixes that (but
possibly not in the most self-evidently correct way).
The test assumes that the current semantics are intended and desired,
which may or may not be true, but it's better than not having any at
all.
* lisp/misc.el (duplicate-line): Don't crash if called with no argument.
* test/lisp/misc-tests.el (misc--duplicate-line): New test.
* lisp/emacs-lisp/bindat.el (bindat--pack-strz): For fixed-length strz
fields, explicitly write a null terminator after the packed string if
there is room (bug#56048).
* doc/lispref/processes.texi (Bindat Types): Update documentation.
* test/lisp/emacs-lisp/bindat-tests.el (bindat-test--str-strz-prealloc):
Update tests.
* lisp/emacs-lisp/bytecomp.el (byte-compile-file): Document behavior
if 'no-byte-compile' is set.
* test/lisp/emacs-lisp/bytecomp-tests.el
(byte-compile-file/no-byte-compile): New unit test.
* test/lisp/emacs-lisp/bytecomp-resources/no-byte-compile.el: New test
file.
* lisp/help.el (substitute-command-keys): Respect 'no-face' argument
also in literal key substitutions.
* test/lisp/help-tests.el
(help-tests-substitute-key-bindings/help-key-binding-face): Rename
from help-tests-substitute-key-bindings/face-help-key-binding.
(help-tests-substitute-key-bindings/help-key-binding-no-face): New test.
* doc/lispref/keymaps.texi (Scanning Keymaps): Document it.
* lisp/keymap.el (make-non-key-event): New function.
* lisp/term/common-win.el (x-setup-function-keys): Mark ns events
as not being keys (bug#55940).
* src/keymap.c (Fwhere_is_internal): Filter out key sequences that
are marked as being non-keys.
Constructs such as ?\C-^@ or ?\C-\C-m literally apply a Control
modifier twice which doesn't make sense at all. What is really meant
is a C0 base character with the Control modifier bit set.
This change is only stylistic in nature.
* lisp/edmacro.el (edmacro-format-keys):
* lisp/keymap.el (key-parse):
* lisp/subr.el (event-modifiers, event-basic-type):
* test/lisp/subr-tests.el (subr-test-kbd):
Use \0 and \r instead of ^@ and \C-m to represent NUL and RET
when combined with other modifiers.
* lisp/emacs-lisp/subr-x.el (hash-table-keys, hash-table-values):
Omit the reversal of the returned list. It is not ordered anyway.
* test/lisp/emacs-lisp/subr-x-tests.el
(subr-x--hash-table-keys-and-values): New test.
Doc strings, `declare` and `interactive` forms must appear in that
order and at most once each. Complain if they don't, instead of
silently ignoring the problem (bug#55905).
* lisp/emacs-lisp/byte-run.el (byte-run--parse-body)
(byte-run--parse-declarations): New.
(defmacro, defun): Check for declaration well-formedness as
described above. Clarify doc strings. Refactor some common code.
* test/lisp/emacs-lisp/bytecomp-resources/fun-attr-warn.el:
* test/lisp/emacs-lisp/bytecomp-tests.el (bytecomp-fun-attr-warn):
New test.
* lisp/emacs-lisp/bindat.el (bindat--pack-strz): Explicitly write a
null byte after packing a variable-length string to ensure proper
termination when packing to a pre-allocated string.
* doc/lispref/processes.texi (Bindat Types): Update documentation.
* test/lisp/emacs-lisp/bindat-tests.el (bindat-test--str-strz-prealloc):
Update tests.
* test/src/timefns-tests.el (decode-then-encode-time)
(time-equal-p-NaN-NaN, time-arith-tests):
Don’t test time functions with infinities and NaNs,
which are not Lisp time values.
Do not assume that the TZ environment variable is either unset or
agrees with /etc/localtime. I ran into this test bug while
running ‘TZ=America/Chicago make check’ on a platform where
/etc/localtime was America/Los_Angeles.
* test/lisp/calendar/icalendar-tests.el:
(icalendar-tests--decode-isodatetime): Don’t use
set-time-zone-rule, whose doc string says “Instead of calling this
function, you typically want something else” for a reason.
Instead, pass the zone arg to icalendar-test--format.
Early normalisation of setq during macroexpand-all allows later
stages, cconv, byte-opt and codegen, to be simplified and duplicated
checks to be eliminated.
* lisp/emacs-lisp/macroexp.el (macroexp--expand-all):
Normalise all setq forms to a sequence of (setq VAR EXPR).
Emit warnings if necessary.
* lisp/emacs-lisp/cconv.el (cconv-convert, cconv-analyze-form):
* lisp/emacs-lisp/byte-opt.el (byte-optimize-form-code-walker):
* lisp/emacs-lisp/bytecomp.el (byte-compile-setq):
Simplify.
* test/lisp/emacs-lisp/bytecomp-tests.el: Adapt and add tests.
* test/lisp/emacs-lisp/bytecomp-resources/warn-variable-setq-nonvariable.el;
* test/lisp/emacs-lisp/bytecomp-resources/warn-variable-setq-odd.el:
New files.
