| Age | Commit message (Collapse) | Author |
|---|
|
Currently, error messages for (syntax) errors do not contain any line
number information. Add the file name and line number to allow users for
easily locating any errors in the corresponding data files.
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
|
|
Conflicts:
src/day.c
src/recur.c
|
|
We did not setup a thread cleanup procedure which resulted in calcurse
freezing if a thread tried to draw on the screen after another thread
was canceled while locking the screen.
Note that this kind of cleanup handlers should be added to other mutexes
as well. This patch just removes the most common case of triggering a
deadlock.
Also note that we cannot move pthread_cleanup_push() and
pthread_cleanup_pop() into the locking/unlocking functions since both
pthread_cleanup_push() and pthread_cleanup_pop() may be implemented as
macros that must be used in pairs within the same lexical scope.
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
|
|
Lock the screen if either the calendar panel or the notification bar is
updated to avoid race conditions.
Addresses BUG#6.
Note that we currently always use a screen-level lock, even if only one
window is affected. This is to be changed in the future.
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
|
|
This is kind of useless since users are not able to react to an
auto-save notification. It also causes all kinds of problems if the
status bar is in a non-standard mode (message, prompt, ...) and is prone
to race conditions if the status bar is updated by another thread at the
same time.
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
|
|
Since the result of localtime() is stored in a statically allocated
structure, data was overwritten when a context switch occurred during
(or shortly after) the execution of localtime(), potentially resulting
in critical data corruption. BUG#7 and BUG#8 are likely related.
This patch converts all usages of localtime() with localtime_r(), which
is thread-safe.
Reported-by: Baptiste Jonglez <baptiste@jonglez.org>
Reported-by: Erik Saule <esaule@bmi.osu.edu>
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
|
|
Remove the export bar and use a simple status_ask_choice() dialog
instead. This reduces code complexity and replaces another unnecessary
menu by the well-tested dialog feature that is used everywhere else.
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
|
|
This is no longer needed since different keys use different handlers
now.
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
|
|
Supersedes the "key" parameter from following functions:
* key_generic_prev_day()
* key_generic_next_day()
* key_generic_prev_week()
* key_generic_next_week()
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
|
|
Split up key_change_priority() into key_raise_priority(),
key_lower_priority() which are wrappers around a new convenience
function change_priority().
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
|
|
Split out key_generic_goto_today() into a separate function. This
simplifies the control flow and supersedes the "key" parameter in
key_generic_goto().
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
|
|
Instead of passing a key, pass the number of steps to increase the
priority by.
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
|
|
Break up the big switch statement into several functions. The functions
are declared inline to avoid unnecessary function calls.
Also:
* Add a "KEY_HANDLER" macro can be used as a short version of a "case"
statement, followed by a function call.
* Declare several state variables global to avoid having to pass them to
each of the new handlers.
* Pass the pressed key to each of the handler functions in order not to
break handlers that handled more than one key before the handlers were
pulled out.
These three changes were introduced to keep this patch as atomic as
possible and are intended to be removed in following patches.
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
|
|
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
|
|
We don't need floating point precision if results are casted back to
integer. Instead, rearrange operations and do the integer division after
the multiplication.
Version numbers are terminating decimals anyway and can be stored using
two integers without losing any information.
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
|
|
Use "p" as an additional (vim-style) default binding for generic-paste.
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
|
|
This is no longer needed. Note removal, as well as exception handling,
have been moved to separate functions and the cut feature has been
merged into the deletion function.
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
|
|
Remove the cut function and merge it into the del-item command. This
allows for vim-style cutting/pasting.
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
|
|
Remove the exception handling code from recur_*_erase() and move it to
separate functions recur_*_add_exc(). Create a wrapper function
day_item_add_exc() that can be used to add an exception to generic
items.
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
|
|
Remove the note removal code from *_delete()/*_erase() and create a new
wrapper function called day_item_erase_note() that is be used to drop
the note being associated to an item.
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
|
|
This adds support for vim-style copy/paste registers which allows
cutting and copying multiple items without having to overwrite the
copy/paste buffer. Registers can be specified using the quote key ('"').
To access a register, type '"x' before a command where "x" is the name
of a register. If you want to copy the currently selected item into
register 1, type '"1c'.
Valid registers are 0-9, a-z, "-" and "_". Note that the latter is the
so-called black hole register, which works similar to the black hole
register in vim.
The register prefix key is currently hardcoded and cannot be configured.
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
|
|
This finally adds full copy-paste support. Implements FR#15.
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
|
|
This can be used to copy an item, so that it can be pasted somewhere
else later.
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
|
|
The "c" key will be used for the generic-copy command in the future. The
configuration menu isn't needed a lot, so binding it to "C" only seems
legit.
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
|
|
This allows pasting items more than once.
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
|
|
Add a new function that can be used to copy one day item into another,
cloning the actual item that is linked.
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
|
|
* Actually duplicate an item instead of copying data only.
* Properly clone an item without a note.
* Mark *_dup() public.
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
|
|
Pass an item instead of passing a list item index (analogous to commit
02c90ba53a658686bad5cb5f88c555d9eef06399).
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
|
|
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
|
|
This is particularly useful if one wants to use configuration and key
bindings from a specific directory, while using an appointment file from
somewhere else. "-c" has precedence over "-D".
Also update the usage message, man page and documentation.
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
|
|
Add a very simple cache, which is used to store the days that contain an
event or an appointment. This makes redrawing and browsing the calendar
panel much faster.
The cache has a size of 31 integers (which is equivalent to 124 bytes on
a 32 bit system and 248 bytes on a 64 bit system) and invalidates itself
if the current month has changed. If an item is added/changed/removed,
the cache needs to be invalidated manually by calling
calendar_monthly_view_cache_set_invalid(). Note that this will always
invalidate the whole cache, even if only one item at the last day of the
month was removed. This is a trade-off between simplicity and
efficiency.
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
|
|
The item type parameter/return value no longer needs to be
passed/returned to interact_day_item_{paste,cut}(), since the type
descriptor is saved when cutting an item now.
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
|
|
Note that this doesn't remove *_dup() since these might still be needed
later.
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
|
|
Instead of calling type-specific duplication handlers and inserting
clones of the original items when pasting, save the generic day item and
remove the actual item from the linked list, so that it can be inserted
anywhere else later.
The cut/paste buffer is moved to the interaction unit, item-specific cut
operations are changed to remove the item from the linked list only
instead of copying and freeing it. An item is only freed if another item
is cut before the current cut/paste buffer is pasted. All paste
operations are changed and reinsert the actual item instead of creating
a clone.
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
|
|
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
|
|
These functions get the current selection, call day_*_item() and fix the
current selection on the appointment panel, so move them where they
belong.
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
|
|
This is no longer used and removing it saves a few bytes per item.
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
|
|
These functions operate on arbitrary items. Pull out the code that gets
the currently selected item, get the current selection when one of the
functions is called and pass it as a parameter.
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
|
|
These functions operate on arbitrary items. Pull out the code that gets
the currently selected item, get the current selection when one of the
functions is called and pass it as a parameter.
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
|
|
Always pass an item instead of passing a date and a index. This makes
use of the NULL callback that was added with one of the previous
patches.
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
|
|
If a NULL callback is passed to llist_find_*(), fall back to comparing
data pointers. The check for a NULL callback pointer is done outside the
main loop with an eye towards performance.
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
|
|
Change the data type of the "data" parameter from "long" to "void *" in
llist_find_*() signatures to allow for passing more complex objects.
Change all llist_find_*() invocations and callbacks accordingly.
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
|
|
Rename all interaction functions for the sake of consistency and add a
"interact_" prefix. Also, mark day_erase_item() static since it is not
used anywhere else.
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
|
|
This is a first step to clean up several compilation units and separate
the front end from back-end operations. All functions that require user
interaction are moved to a new compilation unit "interaction.c". Also,
following things are adjusted to the new layout:
* Make day_item_get_*() and a few other functions public, so that it can
be accessed from the new compilation unit.
* Use apoint_hilt(), todo_hilt(), etc. instead of directly accessing
static variables.
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
|
|
Remove apoint_get(), event_get(), recur_get_apoint() and
recur_get_event(), since they are no longer used.
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
|
|
Pull out code from apoint_switch_notify() -- especially the fallback
routine that calls recur_apoint_switch_notify() if the currently
selected item is a recurrent item -- and move it to a wrapper function
day_item_switch_notify().
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
|
|
Return the total number of items found instead of the pad length, since
this is used to store items in non-interactive mode as well now. The pad
length can still be calculated by using the appointment and event count
parameters.
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
|
|
This kills the hackish code we used to filter and display appointments
and events in non-interactive mode. From now on, the same algorithm that
is used in interactive mode (read day_store_items() for details) is
called, resulting in code that is much easier to maintain.
The resulting performance loss is rather small. Here are the run times
of `calcurse -s01/01/1902 -r18250 -D ../test/data >/dev/null` before and
after the patch was applied:
0.30user 0.26system 0:01.22elapsed 46%CPU (0avgtext+0avgdata 1340maxresident)k
0inputs+0outputs (0major+398minor)pagefaults 0swaps
0.33user 0.21system 0:01.18elapsed 46%CPU (0avgtext+0avgdata 1360maxresident)k
0inputs+0outputs (0major+395minor)pagefaults 0swaps
Note that this also fixes a few bugs that were encountered in
non-interactive mode, most notably BUG#2.
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
|
|
This function allows for writing stored items to stdout in a fashion
similar to day_write_pad(). This will be used as a convenience wrapper
when switching to day_store_items() in non-interactive mode and also
allows for easily implementing a "pipe-day" feature.
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
|
|
Remove the "static" keyword from day_store_items(), so that it is
accessible from other compilation units. Also, allow for discarding the
event/appointment counters by passing NULL pointers and move the
"regex.h" header inclusion to "calcurse.h".
Signed-off-by: Lukas Fleischer <calcurse@cryptocrack.de>
|
