{"thread":{"id":"4331","subject":"What's in git.git","startedAt":"2006-05-29T06:44:06Z","lastAt":"2006-05-29T06:44:06Z","messageCount":1,"participants":["Junio C Hamano"],"isPatch":false,"patchVersion":null,"patchTotal":null},"messages":[{"id":"20867","messageId":"7vac919vpl.fsf@assigned-by-dhcp.cox.net","threadId":"4331","inReplyTo":null,"subject":"What's in git.git","fromName":"Junio C Hamano","fromEmail":"junkio@cox.net","sentAt":"2006-05-29T06:44:06Z","receivedAt":"2006-05-29T06:44:06Z","isPatch":false,"sender":{"key":"gitster@pobox.com","avatar":"https://avatars.githubusercontent.com/u/54884?v=4"},"body":"* The 'master' branch has these since the last announcement.\n  This is a fairly big update.\n\n - git-apply takes notice of beginning and end of file as\n   context (Catalin Marinas, Linus and me)\n      apply: treat EOF as proper context.\n      apply: force matching at the beginning.\n\n - gitview updates (Aneesh Kumar K.V)\n\n - git-mailinfo updates (Eric W. Biederman and me)\n      Make read_one_header_line return a flag not a length.\n      Move B and Q decoding into check header.\n      Refactor commit messge handling.\n      In handle_body only read a line if we don't already have one.\n      More accurately detect header lines in read_one_header_line\n      Allow in body headers beyond the in body header prefix.\n      mailinfo: skip bogus UNIX From line inside body\n      mailinfo: More carefully parse header lines in read_one_header_line()\n\n - format-patch --start-number (Johannes Schindelin and me)\n      cache-tree: replace a sscanf() by two strtol() calls\n      Fix crash when reading the empty tree\n      git-format-patch --start-number <n>\n      built-in format-patch: various fixups.\n      format-patch: -n and -k are mutually exclusive.\n\n - ls-remote rsync:// fix (me)\n      ls-remote: fix rsync:// to report HEAD\n\n   * This should fix clone over rsync:// (which is deprecated\n     but anyway).\n\n - cache-tree optimization for apply & write-tree sequence (me,\n      Johannes, Dennis Stosberg)\n      read-cache/write-cache: optionally return cache checksum SHA1.\n      Add cache-tree.\n      Update write-tree to use cache-tree.\n      Invalidate cache-tree entries for touched paths in git-apply.\n      Use cache-tree in update-index.\n      Add test-dump-cache-tree\n      cache-tree: protect against \"git prune\".\n      index: make the index file format extensible.\n      Teach fsck-objects about cache-tree.\n      cache-tree: sort the subtree entries.\n      test-dump-cache-tree: report number of subtrees.\n      update-index: when --unresolve, smudge the relevant cache-tree entries.\n      read-tree: teach 1 and 2 way merges about cache-tree.\n      read-tree: teach 1-way merege and plain read to prime cache-tree.\n      cache_tree_update: give an option to update cache-tree only.\n      test-dump-cache-tree: validate the cached data as well.\n      cache-tree.c: typefix\n      fsck-objects: mark objects reachable from cache-tree\n      Fix test-dump-cache-tree in one-tree disappeared case.\n      read-tree: invalidate cache-tree entry when a new index entry is added.\n      cache-tree: a bit more debugging support.\n      fsck-objects: do not segfault on missing tree in cache-tree\n      fix git-write-tree with cache-tree on BE64\n\n    * I've held onto this too long but haven't seen breakage.\n      This should make cycles of \"apply & write-tree\" faster by\n      15-20%.\n\n - documentation (Jeff King)\n      cat-file: document -p option\n\n - cvsimport Perl backward compatibility (Jeff King)\n      cvsimport: avoid \"use\" with :tag\n\n - build fixes (Jim Meyering, Martin Waitz)\n      Don't write directly to a make target ($@).\n      Documentation/Makefile: remove extra /\n      Add instructions to commit template.\n\n - \"git-clone --template\" (me)\n      Let git-clone to pass --template=dir option to git-init-db.\n\n - various fixes and cleanups (Linus, Martin Waitz, Petr Baudis,\n   Shawn Pearce, me)\n      builtin-rm: squelch compiler warnings.\n      fetch.c: remove an unused variable and dead code.\n      git-fetch: avoid using \"case ... in (arm)\"\n      bogus \"fatal: Not a git repository\"\n      t1002: use -U0 instead of --unified=0\n      Fix \"--abbrev=xyz\" for revision listing\n      Don't use \"sscanf()\" for tree mode scanning\n      Call builtin ls-tree in git-cat-file -p\n      Built git-upload-tar should be ignored.\n\n - rev-list --objects memory leak fix (Linus)\n      Fix memory leak in \"git rev-list --objects\"\n\n - cvsexportcommit fixups (Yann Dirson)\n      Do not call 'cmp' with non-existant -q flag.\n      Document current cvsexportcommit limitations.\n      Make cvsexportcommit create parent directories as needed.\n\n\n* The 'next' branch, in addition, has these.\n\n - portability of tests across different bourne flavors (Eric Wong)\n      t3300-funny-names: shell portability fixes\n      tests: Remove heredoc usage inside quotes\n      t5500-fetch-pack: remove local (bashism) usage.\n      t6000lib: workaround a possible dash bug\n\n   * I think these are OK to push out to \"master\".\n\n - read-tree/write-tree --prefix from bind commit series (me)\n      read-tree: --prefix=<path>/ option.\n      write-tree: --prefix=<path>\n      read-tree: reorganize bind_merge code.\n\n   * I think these are OK to push out to \"master\".\n\n - avoid wasted work in fetch-pack when receiving side has more\n   roots than the sender (me).\n      fetch-pack: give up after getting too many \"ack continue\"\n\n   * While this would not hurt, it is a client-side hack.  To\n     fix the problem properly, the server side needs to become a\n     bit smarter.\n\n - tree parser reorganization (Linus)\n      Add raw tree buffer info to \"struct tree\"\n      Make \"tree_entry\" have a SHA1 instead of a union of object pointers\n      Switch \"read_tree_recursive()\" over to tree-walk functionality\n      Remove \"tree->entries\" tree-entry list from tree parser\n\n   * This looks good; I would like to cook this for a while in\n     \"next\", and mark its graduation with 1.4.0 release.\n\n - test fix for http-fetch segfault (Sean Estabrooks)\n\n   * Status?\n\n - ref-log (Shawn Pearce)\n      Improve abstraction of ref lock/write.\n      Convert update-ref to use ref_lock API.\n      Log ref updates to logs/refs/<ref>\n      Support 'master@2 hours ago' syntax\n      Fix ref log parsing so it works properly.\n      General ref log reading improvements.\n      Added logs/ directory to repository layout.\n      Force writing ref if it doesn't exist.\n      Log ref updates made by fetch.\n      Change 'master@noon' syntax to 'master@{noon}'.\n      Correct force_write bug in refs.c\n      Change order of -m option to update-ref.\n      Include ref log detail in commit, reset, etc.\n      Create/delete branch ref logs.\n      Enable ref log creation in git checkout -b.\n      Verify git-commit provides a reflog message.\n      Test that git-branch -l works.\n\n   * I think these are OK to push out to \"master\" in that it\n     does not seem to cause regression, but I haven't used this\n     change for real work.  Impressions?\n\n\n* The 'pu' branch, in addition, has these.\n\n - $HOME/.gitrc (Petr Baudis)\n      Read configuration also from ~/.gitrc\n\n   * I like this but it breaks the tests big time.  Not \"next\"\n     material yet, unfortunately.\n"}]}