got stage: implicitly unstage when staging the reverse of the staged diff otherwise we end up with a staged empty edit for that file. ok stsp@

check the maximum cached delta size against the correct constant

convert delta cache to a hash table This approach uses more memory but is much faster. To offset the additional memory usage somewhat the cache now stores very small deltas only. However, overall memory usage goes up. Hopefully we will find a way to reduce this later. ok op@

set `ret' in a few places where it was forgotten; ok tracey@

don't pass $ret to test_done on failure when it's known to be zero Otherwise the test directory it's not left in place; ok tracey@

properly swap cached struct pack array elements in got_repo_cache_pack() Avoids clobbering open files for delta base/accumulation, leaking file descriptors, and triggering errors from close(2) during got_repo_close() as we try to close the same file descriptor more than once.

move creation of tempfiles outside of lib/diff.c ok tracey

map delta cache file into memory if possible while writing a pack file with a fix from + ok op@

portable: fix-up sha1.h include -portable doesn't need this directly.

tog: override SIGTERM and SIGINT handlers to avoid ncurses cleanup() handler ok thomas

reduce GOT_PACK_CACHE_SIZE to 32, otherwise it uses too many open files found by tracey

ensure that all open basefd/accumfd get closed in got_repo_close() found by tracey

open temporary files needed for delta application in got_repo_open() This prepares for callers of got_repo_open() that cannot afford to open files in /tmp, such as gotwebd. In a follow-up change, we could ask such callers to pass in the required amount of open temporary files. One consequence is that got_repo_open() now requires the "cpath" pledge promise. Add the "cpath" promise to affected callers and remove it once the repository has been opened. ok tracey

avoid get_delta_chain_max_size() in dump_delta_chain_to_mem()

avoid get_delta_chain_max_size() in dump_delta_chain_to_file()

fix a bug in findwixt() which caused pack files with missing parent commits The 'nskip' variable is supposed to reflect commits which are waiting on the queue and have the 'skip' color. Only increment 'nskip' when adding such commits to the queue. Problem observed with got send -T and a tag pointing to a deleted branch. Test to reproduce the bug written by op@.

use random seeds for murmurhash2 change the three hardcoded seeds to fresh ones generated on demand via arc4random. Suggested/fixed by and ok stsp@

include header

shrink struct got_pack_meta a bit by removing the have_reused_delta flag This flag can be expressed as m->reused_delta_offset != 0 because all deltas in valid pack files will be written at a non-zero offset. We allocate a huge number of these structs during packing, so every little bit helps.

reduce the amount of memory used for caching deltas during deltification With files sorted properly for deltification we produce better deltas but end up consuming more memory and risk running into OpenBSD ulimits during packing. To compensate, reduce the threshold for the amount of delta data we store in memory, spooling more deltas into the cache file. ok op@

store a path hash instead of a verbatim path in pack meta data This reduces memory use by gotadmin pack. The goal is to sort files which share a path next to each other for deltification. A hash of the path is good enough for this purpose and consumes less memory than a verbatim copy of the path. Git does something similar. ok op@

fix paths stored in pack meta data, improving file deltification The old code was broken and stored an empty path or filenames, instead of a repository-relative path. Which means we didn't sort files for deltification as was intended. Fixing this provides much better deltas in large pack files written by gotadmin pack -a. In my test case, pack size changed from 2GB to 1.5GB. ok op@

refactor got_patch / got_worktree_patch_complete let got_patch own fileindex_path and call got_worktree_patch_complete only if got_wokrtree_patch_prepare hasn't failed. suggested by stsp@

got patch: avoid open/sync/close the fileindex over and over again Instead of flushing the fileindex after every patch in the patchfile just reuse the same fileindex and sync it only at the end of the patch operation. This speeds up 'got patch' on large repositories by quite a lot.

inline struct got_object_id in struct got_object_qid Saves us from doing a malloc/free call for every item on the list. ok op@