aboutsummaryrefslogtreecommitdiffstats
Commit message (Collapse)AuthorAgeFilesLines
* fullscreen: bug: quote $args in the eval line.HEADmasterKaz Kylheku2025-04-081-2/+2
| | | | | | * basta.sh (basta.fullscreen): The args that we calculated with the help of printf %q is shell syntax; it must be double quoted when we eval.
* fullscreen: bug: calculate args after removing options.Kaz Kylheku2025-04-011-1/+2
| | | | | | | * basta.sh (basta.fullscreen): We must take the arguments for the program after recognizing and removing the function's options, otherwise we treat those options as part of the command.
* fullscreen: regression: no args in non-fulscreen case.Kaz Kylheku2025-03-191-1/+1
| | | | | | | | * basta.sh (basta.fullscreen): Move the calculation of the args variable to the top. It was not yet set in the case when we run the command without fullscreen and return (because standard output is not a TTY, and -f had not been specified).
* fullscreen: don't fullscreen if stdout is not a tty.Kaz Kylheku2025-03-182-12/+44
| | | | | | | | | | | * basta.sh (basta.fullscreen): Separate option parsing from side effects an do it earlier. Recognize a new -f option, separately or clumped with -s. If -f is not supplied and standard output is not a TTY, then just execute the command and return, without doing full screen. * README.md: Documented -f option.
* fullscreen: use eval to improve appearance of suspended job.Kaz Kylheku2025-03-041-1/+2
| | | | | | | | | | | | | | | | | | | | | | | | There is a difference between Bash 4 and Bash 5. When a command which has been invoked by syntax like: command "$@" is suspended by Ctrl-Z, Bash 4 shows, as the displayed job string, the real program that is running, and its arguents. Whereas Bash 5 just shows the above unexpanded 'command "$@"' syntax! To work around this, we expand "$@" ourselves, and eval an expanded command. * basta.sh (basta.fullscreen): Expand the command arguments into a properly escaped variable with the help of the %q argument of printf. Then we eval the command. We keep the LINES=$realrows out of the command, putting that into the eval, otherwise that will appear in the jobs also. The cosmetics is improved. For instance if the Bash 5 user has an alias like "alias vim='basta.fullscreen vim'", and then runs "vim file" and suspends the editor into the background, they will see "command vim file" as the job name, rather than LINES=$realrows command "$@".
* Gnome Terminal doesn't take line wrap escape sequences.Kaz Kylheku2025-02-171-2/+0
| | | | | | * basta.sh (basta.update_status): Gnome Terminal just prints ESC [ = 7 l rather than executing it. Let's just drop these escapes.
* Get rid of deep flag in update.Kaz Kylheku2025-02-171-19/+12
| | | | | | | | | * basta.sh (basta.update_status): Eliminate the deep flag. Always repaint all the status lines together. Without this, if the user runs the reset command, the lower status lines disappear. I.e. it worked better with the bug whereby we were always setting deep=y.
* bug: unnecessarily deep refresh.Kaz Kylheku2025-02-171-5/+9
| | | | | | | | * basta.sh (basta,update_status): Fix another bogus use of || and && into if statement. We only want deep=y to be evaluated whenever the test fails; i.e. to do the deep repaint in the same case when we called basta.prepare_term.
* fullscreen: code cleanup and minor issue.Kaz Kylheku2025-02-161-2/+4
| | | | | | | | | | | | * basta.sh (basta.fullscreen): Clear to to the end of screen in the else case of testing for -s, and set basta_deferred_intr there. It doesn't matter that we do this clearing before resetting the scrolling region. A a result, we don't have to test the $saveterm variable again before running the command. The old structure with || and && was not correct; the intent is to set basta_deffered_intro to WINCH only if we are not saving and restoring the terminal.
* Minor code cleanup.Kaz Kylheku2025-02-151-3/+2
| | | | | | | * bast.sh (basta.query_terminal_lines): Remove unused curline variable. (basta.fullscreen): Likewise. Also move curly brace to next line.
* README: add Bugs section, similarly to a man page.Kaz Kylheku2025-02-151-0/+15
| | | | * README.md: New section, Bugs.
* Regression: clock not spontaneously updating.Kaz Kylheku2025-02-141-1/+5
| | | | | | | | | | | | | * basta.sh (basta_executing): Don't preserve the value of this variable when reloading. Set it to y, and then blank at the end of the script when loading is about to end. This fixes the bug: we had the variable set to y, and so basta.interrupt was deferring signal handling. (basta.initial_prompt_hook): Bind basta_executing to y while running this funcion also, like basta.prompt_hook. (basta.fullscreen): Also here; while running any command, we don't need to handle ALRM signals, only while editing the command line.
* Simplify fullscreen to one function.Kaz Kylheku2025-02-142-79/+24
| | | | | | | | | | | | | | | | | * basta.sh (badsta.fullscreen); Merges functionality of basta.fullscreen_alt. When -s is not specified, clears to end of screen, and calls basta.ensure_bottom_margin. When calling this function, we use the current value of $LINES rather than recalculating realrows, removing one interaction to the terminal. Since we are in full screen mode, we can trust that $LINES is tracking the size. When -s is specified, we don't clear to end of screen or call basta.ensure_bottom_margin. Also we don't need to provoke WINCH handling. (basta.fullscreen_alt): Function removed. * README.md: Documentation trimmed and revised.
* Share logic between prepare_term and fullscreen_alt.Kaz Kylheku2025-02-141-22/+15
| | | | | | | | | | | | * basta.sh (basta.ensure_bottom_margin): New function, factored off from basta.prepare_term. (basta.prepare_term): Code for ensuring that screen contents move out of status-sized bottom area now moved into basta.ensure_bottom_margin. (basta.fullscreen_alt): Complex logic involving knowing the current line, moving the cursor to the bottom, and scrolling a certain nummber of lines removed. Replaced with call to basta.ensure_bottom_margin.
* prepare_term: account for all status lines.Kaz Kylheku2025-02-131-1/+6
| | | | | | | | | | | | * basta.sh (basta.prepare_term): The old logic of printing \n ESC [ A (newline, go up one line) is inadequate when we have multiple previous status lines to draw. It only works if the previous status lines can be counted on to be already scroll-protected. We need to print as many new lines as we have status lines, and then move the cursor up that many lines. Also, before doing this, we reset the scrolling region to the full size. This same logic could simplify basta.fullscreen_alt.
* basta.fullscreen_alt: fix off-by-one bug in scroll logic.Kaz Kylheku2025-02-131-2/+2
| | | | | | | | | | * basta.sh (basta.fullscreen_alt): The worst case number of scrolls needed is basta_prev_reserved_rows + 1, the height of the status area, not not basta_prev_reserved_rows. Now the [ $scrolls_needed -ge 0 ] test is correct, and we don't we don't need the extra newline in the final printf. The behavior is now good in the cases when the command prints zero, one, two, and three lines.
* get_cur_line: don't clobber var on failure.Kaz Kylheku2025-02-131-5/+3
| | | | | | | | | | | | | * basta.sh (basta.query_terminal_lines): Streamline by passing received variable name down to get_cur_line, avoding intermediate variable. We don't need a nameref now since we are just passing the name down. (basta.get_cur_line): Do not assign into the variable and test for validity after; test "$2" for validity and only if valid assign to variable. This is taken advantage in basta.fullscreen_alt where the original estimate $realrows is unaffected if query_terminal_lines fails.
* basta.fullscreen_alt: assume worst case if no realowsKaz Kylheku2025-02-131-3/+9
| | | | | | | * basta.sh (basta.fullscreen_alt): If not able to obtain the current line and number of terminal lines from the terminal, then assume that the worst case number of scrolls are needed: size of the status area.
* Return and check status from get_cur_line.Kaz Kylheku2025-02-131-9/+14
| | | | | | | | | | | | | | | | | From time to time get_cur_line doesn't fetch a value due to a timeout or, more likely, interference from TTY input. We have to check those situations so as not to try to calculate with a bad value or pass it to stty or whatever. * basta.sh (basta.query_terminal_lines): Propagate status now returned by get_cur_line. (basta.prepare_term): Skip logic and return failure if query_terminal_lines failed, since realrows contains garbage. (basta.get_cur_line): Return success if value obtained is a string of one or more decimal digits. (basta.check_cursor): Improve error checking. (basta.fullscreen_alt): Error checks.
* bug: fullscreen interactions with completion.Kaz Kylheku2025-02-131-0/+10
| | | | | | | | | | | | | | | | When the Bash completion stuff for git is loaded, it generates a __git() function. That function's body calls the user's expanded alias, such as "basta.fullscreen_alt git". This expansion of the alias is baked into that function's code. Other completion modules might do a similar thing. The most prudent thing we can do is to assume that calls to basta.fullscreen or basta.fullscreen_alt can happen during tab completion, and in those cases just run the command and bail without doing any full screen logic. * basta.sh (basta.fullscreen, basta.fullscreen_alt): If the $COMP_WORDS variable is not empty, run the command and bail.
* fullscreen: bug: must redirect helpers to tty.Kaz Kylheku2025-02-131-2/+2
| | | | | | | * basta.sh (basta.fullscreen_alt): When calling query_terminal_lines and get_cur_line, we must redirect the input and output to $ttyfd, like we do for our local printf and stty calls.
* Use /dev/tty rather than disabling fullscreen.Kaz Kylheku2025-02-131-20/+20
| | | | | | | | | | | | | | | In the full screen functions, our strategy of disabling fullscreen mode when standard input or output are redirected is suboptimal. What we should be doing is opening /dev/tty and doing our terminal interactions with that, and not care that standard input and output are redirected. * basta.sh (basta.fullscreen, basta.fullscreen_alt): Remove the check for stdin and stdout not being a terminal, which runs the command without fullscreen mode. Open /dev/tty for reading and writing, retaining the file descriptor in the ttyfd variable. Redirect all tty interactions to this ttyfd descriptor.
* Add terminal save-restore feature to basta.fullscreen.Kaz Kylheku2025-02-122-8/+42
| | | | | | | | | | * README.md: Revised full screen documentation. * basta.sh (basta.fullscreen): Recognize and consume a -s option. If this is present, then issue the ANSI escape sequence for saving the terminal before running the command. Then issue the sequence for restoring the terminal afterward.
* fullscreen: handle stdin/stdout not being a TTY.Kaz Kylheku2025-02-121-0/+10
| | | | | | | | | | | | * basta.sh (basta.fullscreen, basta.fullscreen_alt): Only perform the special fullscreening logic if the standard input and output are a TTY. If either one is not a TTY, then just run the command and return its status. Otherwise if the program is redirected, the codes we send to the terminal get mixed up in the output. This test isn't perfect, because programs could bre redirected to a TTY which is not the controlling TTY. This is rare to do, and users probably won't be doing this with their basta.fullscreen aliases.
* Don't unexport entire status array in basta.cleanup.Kaz Kylheku2025-02-121-11/+3
| | | | | | | | * basta.sh (basta.cleanup): What we actually want to do is restore the previous status line environment that we inherited. We do this by deleting the entry that we added to the top of the array and re-exporting the array. (basta.unexport_array): Unused function removed.
* Unexport the status line array in basta.cleanup.Kaz Kylheku2025-02-121-0/+1
| | | | | | | | | | * basta.sh (basta.cleanup): Unexport the LC_basta_status array. If we don't do this, then when the user performs the recommended deep recycle on Basta, consisting of manually invoking basta.cleanup followed by exec bash, the user will see an extra status line. The executed program sees the environment published by the previous instance, which Basta interprets as status line recursion.
* Better handling of SIGINT and failed status in fullscreen.Kaz Kylheku2025-02-111-4/+21
| | | | | | | | | | | * basta.sh (basta.prompt_hook): Exchange the order of update_status and do_exit_status. When we are coming out of fullscreen, if we print the exit status, it ends up in the wrong place due to how we moved the cursor. (basta.fullscreen, basta.fullscreen_alt): Disable SIGINT around most of the function so the command is interrupted, but the function isn't.. Capture the exit status and return it. This way the function doesn't bail.
* Add some reentrancy discipline to the interrupts.Kaz Kylheku2025-02-111-4/+22
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | - A basta_executing flag is introduced. The ALRM and WINCH interrupts will not execute basta.update_status if the flag is true (nonblank). The flag is false while the command line is being edited. - When an interrupt finds the basta_executing true, it stores its keyword into basta_deferred_intr. Code later is then aware that the interrupt happened and can take some action. We take advantage of this to cancel the holdoff time in basta.check_cursor, and in the fullscreen functions to fake a WINCH, which basta.check_cursor and basta.prompt_hook will react to. Launching an an editor in full screen and quickly exiting it is now perfectly reliable; the cursor deosn't briefly end up in the status area. * basta.sh (basta_executing, basta_deferred_intr): New global variables. (basta.check_cursor): Ignore the 3 second hold-off period if a deferred interrupt has been recorded. We could probably specifically check for it being WINCH here, but this is fine. (basta.prompt_hook): Initialize the winch variable from basta_deferred_intr, so that we will force a fulle rupdate if a WINCH interrupt balked. Set the basta_executing flag over the execution of the update logic. Clear basta_deferred_intr, but not after basta.check_cursor has had a chance to look at it. (basta.interrupt): New function: the interrupt hooks now point to this function. (basta.install_hooks): Point ALRM and WINCH traps to basta.interrupt. (basta.fullscreen, basta.fullscreen_alt): Pretend that a WINCH interrupt was just deferred, so the update logic in basta.prompt_hook will react as described above, to expedite the cursor out-of-bounds :heck, and update of the status area.
* Do not save and restore screen for wrap mode.Kaz Kylheku2025-02-121-2/+2
| | | | | | | | * basta.sh (basta.update_status): When repainting the parent status lines, just disable wrap mode and enable it again. Saving and restoring the terminal is doing weird things in some terminals, like very quickly flashing some previously saved screen.
* Bungled $'...' literal.Kaz Kylheku2025-02-121-1/+1
| | | | | * basta.sh (basta.fullscreen_alt): Fix transposed ' and $ in $'...' literal for clearing to end of screen.
* Improve the basta.fullscreen_alt mode.Kaz Kylheku2025-02-112-5/+11
| | | | | | | | | | | | | * basta.sh (basta.fullscreen_alt): More nicely handle any program. Clear to the end of screen before the program runs to get rid of the status lines, so that the program's output does not get mingled with them. Do the correct amount of scrolling based on the cursor position observed after the program has run, so that the program's last line of output is above the status area, plus prompt space. * README.md: Recommend basta.fullscreen_alt for git, because git doesn't always launch an editor or pager.
* Use environment variables to convey nested status lines.Kaz Kylheku2025-02-112-24/+111
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Starting with this comming, Basta will no longer look at the difference between `$LINES` and the measured terminal size to determine the presence of existing status lines. Basta now maintains an array of status lines. Each recursive instance adds a status line to the array. The array is propagated through the environment via numbered environment variables. The names of the array variables is chosen in the LC_* namespace such that they will propagate across ssh sessions; because Debian, Ubuntu and perhaps other distros configure OpenSSH to pass variables in this namespace. * README.md: Documented and spellchecked. * basta.sh (basta.prepare_term): Do not calculate basta_prev_preserved_rows as the difference between realrows and LINES. Instaed, measure the LC_basta_status array. (basta.update_status): When doing a full refresh, repaint the status lines provided to us from any parent Basta sessions. We don't try to fit these to the terminal width; instead we put the terminal in non-wrapping mode temporarily. Also, when calculating the current (topmost) status line, stick it into the topmost position in the array and update the environment variables from the array. (basta.check_cursor): Small off-topic change here. We adjust the holdoff time from 5 seconds to 3 seconds. I'm finding 5 to be a little too slow. (basta.export_array, basta.import_array, basta.unimport_array): New functions. (basta.fullscreen, basta.fullscreen_alt): remove the hack of exporting an incremented value of basta_prev_reserved_rows into the environment. No such thing is now required for any nested Basta session run out of an editor to reserve its status line above existing status lines. In basta.fullscreen_alt we also fix a variable name typo.
* Improve basta.fullscreen; add basta.fullscreen_alt.Kaz Kylheku2025-02-102-2/+46
| | | | | | | | | | | | - basta.fullscreen sets LINES also in addition to stty rows, and also resets the scrolling region. - bata.fulscreen_alt uses an escape sequence to move to the last line instead of multiple \n characters. Additionally, after moving to the last line, it issues some newlines to scroll the screen, so that any material printed by the executed program is out of the way of the status area (taking into account stacked status lines).
* Refactor: basta.get_cur_line now puts value into variable.Kaz Kylheku2025-02-101-3/+7
|
* New feature: basta.fullscreen.Kaz Kylheku2025-02-102-1/+24
| | | | | | | - When you invoke `basta.fullscreen command [arg ...]`, the external command `command` will make full use of the terminal. This can be set up as an alias for certain commands, like your favorite text editor.
* README: additional installation notes.Kaz Kylheku2025-02-101-0/+31
| | | | | | - Some advice about Bash configuration for people who are not Bash experts. Bash trips people up by reading different config files when run as a login shell versus non-login shell.
* Don't interrogate terminal too often in basta.check_cursor.Kaz Kylheku2025-02-101-9/+30
| | | | | | | | - We take advantage of the $EPOCHSECONDS variable available in newer Bash to get the current time. We don't perform the cursor position or resize check in basta.check_cursor if it has been done in the last 5 seconds, in addition to continuing to avoid it if there is buffered TTY input.
* Check for changed window size at prompt time.Kaz Kylheku2025-02-101-2/+12
| | | | | | | | | | | | - In basta.check_cursor, we already query the terminal's current position via basta.get_cur_line, and avoid doing so if there is pending TTY input. Why don't we also check the screen size here. If it has changed, we yield a false return status. - In basta.prompt_hook, we check the return status of basta.check_cursor. If it is failed, then we add the WINCH parameter when calling basta.update_status.
* Fix one remaining instance of \033 escape to $'...' and \e.Kaz Kylheku2025-02-061-1/+1
|
* Distinguish WINCH in basta.update_status.Kaz Kylheku2025-02-061-2/+4
| | | | | | | | | | | | | | | | - When SIGWINCH and SIGALRM are received and we are calling basta.update_status because of that, pass in an argument which allows that to be distinguished. basta.update_status will call basta.prepare_term if it was invoked because of SIGWINCH, even the apparent terminal size has not changed. - One upshot of this is that the user can recover a broken scrolling region now by doing "kill -WINCH $$", which will work even in the pathological situation when the full screen is scrolling, and basta_scroll_lines, LINES and stty rows all contain the same value (the full screen height).
* Watch for empty response when getting current line.Kaz Kylheku2025-02-031-1/+1
| | | | | | | | - When we query the current line from the terminal, there are situations when that might time out and yield a blank response. In basta.check_cursor, we guard against that, or else the user may, rarely, see a message complaining that the [ command is missing an operand for the -gt operator.
* Do not set termios rows in prompt hooks.Kaz Kylheku2025-01-311-2/+2
| | | | | | | - I suspect this is involved in the bug I'm seeing whereby the scroll region goes wonky. It may be preventing our logic from kicking in to fix the situation, by making termios_lines equal LINES and to basta_old_lines.
* Use \e for ESC via Bash $'...' syntax.Kaz Kylheku2025-01-311-18/+12
| | | | | - We get rid of 'local esc=$(printf "\033")' everywhere and use the escape sequence.
* README: recommend simple reload first.Kaz Kylheku2025-01-261-11/+9
|
* Abruptly disappearing status bug.Kaz Kylheku2025-01-141-1/+1
| | | | | | | | | | | - In the basta.update_status function's path abbreviating loop, we must test whether the path can be abbreviated, and bail, before doing the abbreviation. If we do the test after the abbreviation, we create a one off error where we bail the loop with an overlong status line that does not fit even though we could have abbreviated the path by one more component. And then the status line is surprisingly blanked.
* Copyright year bump to 2025.Kaz Kylheku2025-01-011-1/+1
|
* Check the termios rows and columns for sanity.Kaz Kylheku2024-12-311-1/+18
| | | | | | | | | | | | | | | - In basta.update_status, we should be checking not only for LINES and COLUMNS changing, but also for the underlying termios parameters changing (which we can obtain using stty size). The motivation for this change is that I'm still seeing situations in which the scroll region gets messed up. I can get Basta to fix it by manually setting basta_old_lines to 0 to force basta.update_status to call basta.prepare_term. That tells me that the problem is that basta_old_lines is equal to $LINES. I've not root-caused the issue, though. The hypothesis is that perhaps $LINES has become incorrect: perhaps Bash didn't set LINES to the size that the kernel knows about and has set up in termios. We can catch that discrepancy.
* Streamline code in basta.prepare_term.Kaz Kylheku2024-12-191-4/+4
| | | | | | | | - We calculate the final LINES value unconditionally whether or not it is the first call that sets basta_prev_reserved_rows. Then we just copy that value into basta_scroll_lines and basta_old_lines. We don't rely on the side effect of the "stty rows" command setting LINES.
* bugfix: don't clobber LINES with real terminal sizeKaz Kylheku2024-12-191-3/+4
| | | | | | | | | - In basta.prepare_term, if we already know how many reserved rows there are, we should not be calling bsata.query_terminal_lines to get the real terminal size into LINES. We need to get that into a local variable and then subtract form that the lines we know are reserved, and use that for setting the number of lines.
* bugfix: nameref hygiene problem.Kaz Kylheku2024-12-191-2/+2
| | | | | | | | - the basta.query_terminal_lines function is invoked by a caller which passes in a variable name realrows. If the function has a same-named local variable, the name wrongly refers to that variable. We rename the local to curline.
generated by cgit v1.2.3 (git 2.25.1) at 2025-04-23 08:54:17 +0000