Commit Briefs
Add a refs argument to got_repo_match_object_id(), replacing 'resolve_tags'
Make use of this where possible to avoid re-reading references from disk. ok naddy
make use of reflist object id maps in tog log and diff views
This is a partial fix for performance issues when browsing the FreeBSD Git repo, which can easily contain more than 4000 references. ok naddy
do not mix up repositories if tog's -r option is used inside a work tree
with input from and ok naddy
use size_t for loop indices to avoid signedness warnings; by emaste@freebsd
Same change as 16aeacf7088d, for subdirectories other than lib/
replace unprintable characters with '.' before passing them to curses
Otherwise, ncurses will replace them with some printable representation whose width we can't predict, and wunctrl() fails to return the replacement for 0x80..0x9f. ok stsp
fix display of lines that end in "\r\n" in tog(1)
Problem found and reported by jrick ok naddy
move the ref view's selection cursor up if the terminal shrinks too much
bug found by naddy ok naddy
send resize events to tog child views as well as parent views
problem found by naddy ok naddy
fix crashes when the 'tog log' view reloads displayed data
This reimplements log view reloading (Ctrl-L), logging of a parent path (Backspace), and the toggle to show commits on branches (B). The idea is to reuse the existing log view and change its state, instead of allocating a new view with a new state and replacing the existing view. Fixes a segfault that occurs when a parent path is logged with Backspace: tog tree -r got.git -c 0.44 pick tog/tog.c 'l' Backspace -> tog will segfault The first change in this patch is a partial fix. The log thread should always check the 'quit' flag as soon as it wakes from sleep. Otherwise it could try to load more commits after waking up and before checking the 'quit' flag. It will then attempt to load commits with a NULL commit graph pointer. This partial fix by itself is not sufficient to fix the crash, since we'll now see a bus error in the main thread, instead of a NULL deref in the log thread. The remainder of the patch fixes this bus error. ok naddy