move remaining got_opentemp's out of blame. ok op@

tog: implement support for horizontal splitscreens Enable all main views to open primary child views in a horizontal split: log -> diff tree -> blame blame -> diff ref -> log Support is yet to be expanded to nested child views and key maps for manipulating split type and size. ok op@ and stsp@

tog: refactor log cursor movement in prep for hsplit Simplify the log input handler by moving (page)up/down and home/end logic into helper routines as we need to perform some checks on these cases when navigating the log view in horizontal splits. ok op@ and stsp@

fix split-screen issue in tog when moving focus from child view to parent view Reported by jamsek: $ tog # term wide enough to vsplit return # open commit f # fullscreen commit tab # should go to fullscreen log, but it splits the screen fix ok jamsek

fix recent regression in tog where split view was not restored after resize ok jamesk

move got_opentempfd() out of lib/diff.c again ok tracey

move got_opentempfd out of blame_open ok stsp@ jrick@

move got_opentempfd out of open_blob. ok stsp@

tog: echo count prefix when entering compound key maps like less(1), echo the digits as they're typed. suggested by and ok op@ stsp@

always show commit or tree IDs in diff header, in order to help 'got patch' The idea is that got patch can simply look for a line such as: commit - abcde1234567... to find the merge base commit ID to show in diff3 conflict markers. got log -p now displays commit IDs in its diff header, instead ofl tree or blob IDs. ok op@

build with -Wmissing-prototypes ok stsp@

tog: prefix mvmt keymaps with count modifier like vi(1) Enable compound commands for all movement keymaps (e.g., hjkl, bfdu). The optional prefixed count determines how many times the movement is performed; for example, '123j' (move 123 lines down) or '4f' (scroll 4 pages down). Counts are ignored if preceding non-movement and some singular movement operations (e.g., 't', 'r', 'ret', '/', '$', 'G', 'g'). ok stsp@

fix `limit' accounting in draw_commit if the child view is fullscreen we don't need to subtract one, because no vborder is drawn. ok Mark Jamsek

tog: make SPACE page down in all views ok op@, Mark Jamsek

fix fullscreen view regression introduced in 0dbbbe90fd When cycling from a child view in fullscreen, the parent view is opened in a splitscreen with no border instead of fullscreen. Check the view mode when cycling between views and resize parent dimensions accordingly. ok op@

alias C-b/C-f to scroll page back/forward with b/f Suggested by naddy. Remap 'f'ullscreen to 'F', and blame view key maps 'b' and 'B' to 'c' and 'C', respectively, per stsp's and naddy's suggestion. While here, remove trailing whitespace. ok naddy@

small grammar and accuracy fixes in a comment

add missing braces around multi-line if-statement body; pointed out by op@

take tog selection cursor position into account when searching for next match Patch by Mikhail ok op@ jamsek

tog: reset matched and search data after C-l otherwise they point to free'd data. ok stsp@

fix a segfault in tog diff The f1 tempfile must always be created. Even if the initial diff has no objects on the left side, the view can be switched to a different diff which does have objects on both sides. When that happened, tog crashed. (To reproduce: open tog in got.git, hit G, hit Enter, hit <)

fix build of tog on sparc64 with -Werror error was tog/tog.c:636: warning: function declaration isn't a prototype

account for the vborder in draw_commit the vborder is drawn on top of the window and that can cause issues with double-width characters. Instead, account for it when computing the space available for the log message. ok stsp@

tog: resize main view drawing the child view on top of the main view when the layout is split-screen can produce some ugly glitches when double-width characters are involved. (issue found by stsp) ok stsp@

simplify scroll_wline (now span_wline) and format_line generalize scroll_wline to start at a specified offset so that we can reuse it to simplify format_line. While here, simplify scroll_wline (now called span_wline) too. ok and improvements by stsp@