Commits


extract large objects to a temporary file in got-index-pack


add mmap support to got-index-pack


do not buffer more data than necessary in got_inflate_to_mem_fd()


avoid re-reading the entirety of an object's data to calculate the CRC


rewritten got-index-pack; sorry about the monster commit


add optional 'consumed' output parameter to got_inflate_to_mem_fd()


remove unneeded memory allocations from ref delta parser which were leaking


add support for first-parent history traversal to got-read-pack


add copyright year for files already touched in 2020


fix reading ref deltas with GOT_PACK_NO_MMAP builds


remove obj->path_packfile which is unused nowadays; saves a few free() calls


cache delta data buffers in an LRU cache


stop storing decompressed delta data in struct got_delta; fetch it on demand


when an object id string is bad, include the bad string in the error message


fix reading pack files larger than 2GB; ok tb@; problem found by mpi@


properly handle read errors in got_packidx_init_hdr (found by jj@ scan-build)


plug error leak in got_packidx_init_hdr(); found by jj@ with scan-build


got_packidx_open: free packidx object when open(2) fails


fix reading pack index files with large offsets (patch by Hiltjo Posthuma)


fix build on OpenBSD/sparc64 (gcc)


call got_object_id_queue_free() instead of implementing it inline


disambiguate abbreviated object IDs based on object type


speed up matching of abbreviated commit IDs in pack files


introduce support for abbreviated object IDs


replace got_pack_get_packfile_size() with simple fstat()