Documentation/RelNotes/2.3.0.txt

   1 Git v2.3 Release Notes
   2 ======================
   3
   4 This one ended up to be a release with lots of small corrections and
   5 improvements without big uncomfortably exciting features.  The recent
   6 security fix that went to 2.2.1 and older maintenance tracks is also
   7 contained in this update.
   8
   9
  10 Updates since v2.2
  11 ------------------
  12
  13 Ports
  14
  15  * Recent gcc toolchain on Cygwin started throwing compilation warning,
  16    which has been squelched.
  17
  18  * A few updates to build on platforms that lack tv_nsec,
  19    clock_gettime, CLOCK_MONOTONIC and HMAC_CTX_cleanup (e.g. older
  20    RHEL) have been added.
  21
  22
  23 UI, Workflows & Features
  24
  25  * It was cumbersome to use "GIT_SSH" mechanism when the user wanted
  26    to pass an extra set of arguments to the underlying ssh.  A new
  27    environment variable GIT_SSH_COMMAND can be used for this.
  28
  29  * A request to store an empty note via "git notes" meant to remove
  30    note from the object but with --allow-empty we will store a
  31    (surprise!)  note that is empty.
  32
  33  * "git interpret-trailers" learned to properly handle the
  34    "Conflicts:" block at the end.
  35
  36  * "git am" learned "--message-id" option to copy the message ID of
  37    the incoming e-mail to the log message of resulting commit.
  38
  39  * "git clone --reference=<over there>" learned the "--dissociate"
  40    option to go with it; it borrows objects from the reference object
  41    store while cloning only to reduce network traffic and then
  42    dissociates the resulting clone from the reference by performing
  43    local copies of borrowed objects.
  44
  45  * "git send-email" learned "--transfer-encoding" option to force a
  46    non-fault Content-Transfer-Encoding header (e.g. base64).
  47
  48  * "git send-email" normally identifies itself via X-Mailer: header in
  49    the message it sends out.  A new command line flag --no-xmailer
  50    allows the user to squelch the header.
  51
  52  * "git push" into a repository with a working tree normally refuses
  53    to modify the branch that is checked out.  The command learned to
  54    optionally do an equivalent of "git reset --hard" only when there
  55    is no change to the working tree and the index instead, which would
  56    be useful to "deploy" by pushing into a repository.
  57
  58  * "git new-workdir" (in contrib/) can be used to populate an empty
  59    and existing directory now.
  60
  61  * Credential helpers are asked in turn until one of them give
  62    positive response, which is cumbersome to turn off when you need to
  63    run Git in an automated setting.  The credential helper interface
  64    learned to allow a helper to say "stop, don't ask other helpers."
  65    Also GIT_TERMINAL_PROMPT environment can be set to false to disable
  66    our built-in prompt mechanism for passwords.
  67
  68  * "git branch -d" (delete) and "git branch -m" (move) learned to
  69    honor "-f" (force) flag; unlike many other subcommands, the way to
  70    force these have been with separate "-D/-M" options, which was
  71    inconsistent.
  72
  73  * "diff-highlight" filter (in contrib/) allows its color output to be
  74    customized via configuration variables.
  75
  76  * "git imap-send" learned to take "-v" (verbose) and "-q" (quiet)
  77    command line options.
  78
  79  * "git remote add $name $URL" is now allowed when "url.$URL.insteadOf"
  80    is already defined.
  81
  82  * "git imap-send" now can be built to use cURL library to talk to
  83    IMAP servers (if the library is recent enough, of course).
  84    This allows you to use authenticate method other than CRAM-MD5,
  85    among other things.
  86
  87  * "git imap-send" now allows GIT_CURL_VERBOSE environment variable to
  88    control the verbosity when talking via the cURL library.
  89
  90  * The prompt script (in contrib/) learned to optionally hide prompt
  91    when in an ignored directory by setting GIT_PS1_HIDE_IF_PWD_IGNORED
  92    shell variable.
  93
  94
  95 Performance, Internal Implementation, Development Support etc.
  96
  97  * Earlier we made "rev-list --object-edge" more aggressively list the
  98    objects at the edge commits, in order to reduce number of objects
  99    fetched into a shallow repository, but the change affected cases
 100    other than "fetching into a shallow repository" and made it
 101    unusably slow (e.g. fetching into a normal repository should not
 102    have to suffer the overhead from extra processing).  Limit it to a
 103    more specific case by introducing --objects-edge-aggressive, a new
 104    option to rev-list.
 105
 106  * Squelched useless compiler warnings on Mac OS X regarding the
 107    crypto API.
 108
 109  * The procedure to generate unicode table has been simplified.
 110
 111  * Some filesystems assign filemodes in a strange way, fooling then
 112    automatic "filemode trustability" check done during a new
 113    repository creation.  The initialization codepath has been hardened
 114    against this issue.
 115
 116  * The codepath in "git remote update --prune" to drop many refs has
 117    been optimized.
 118
 119  * The API into get_merge_bases*() family of functions was easy to
 120    misuse, which has been corrected to make it harder to do so.
 121
 122  * Long overdue departure from the assumption that S_IFMT is shared by
 123    everybody made in 2005, which was necessary to port to z/OS.
 124
 125  * "git push" and "git fetch" did not communicate an overlong refname
 126    correctly.  Now it uses 64kB sideband to accommodate longer ones.
 127
 128  * Recent GPG changes the keyring format and drops support for RFC1991
 129    formatted signatures, breaking our existing tests.
 130
 131  * "git-prompt" (in contrib/) used a variable from the global scope,
 132    possibly contaminating end-user's namespace.
 133
 134
 135 Also contains various documentation updates and code clean-ups.
 136
 137
 138 Fixes since v2.2
 139 ----------------
 140
 141 Unless otherwise noted, all the fixes since v2.2 in the maintenance
 142 track are contained in this release (see the maintenance releases'
 143 notes for details).
 144
 145  * "git http-push" over WebDAV (aka dumb http-push) was broken in
 146    v2.2.2 when parsing a symbolic ref, resulting in a bogus request
 147    that gets rejected by recent versions of cURL library.
 148    (merge f6786c8 jk/http-push-symref-fix later to maint).
 149
 150  * The logic in "git bisect bad HEAD" etc. to avoid forcing the test
 151    of the common ancestor of bad and good commits was broken.
 152    (merge 07913d5 cc/bisect-rev-parsing later to maint).
 153
 154  * "git checkout-index --temp=$target $path" did not work correctly
 155    for paths outside the current subdirectory in the project.
 156    (merge 74c4de5 es/checkout-index-temp later to maint).
 157
 158  * The report from "git checkout" on a branch that builds on another
 159    local branch by setting its branch.*.merge to branch name (not a
 160    full refname) incorrectly said that the upstream is gone.
 161    (merge 05e7368 jc/checkout-local-track-report later to maint).
 162
 163  * With The git-prompt support (in contrib/), using the exit status of
 164    the last command in the prompt, e.g.  PS1='$(__git_ps1) $? ', did
 165    not work well, because the helper function stomped on the exit
 166    status.
 167    (merge 6babe76 tf/prompt-preserve-exit-status later to maint).
 168
 169  * Recent update to "git commit" broke amending an existing commit
 170    with bogus author/committer lines without a valid e-mail address.
 171    (merge c83a509 jk/commit-date-approxidate later to maint).
 172
 173  * The lockfile API used to get confused which file to clean up when
 174    the process moved the $cwd after creating a lockfile.
 175    (merge fa137f6 nd/lockfile-absolute later to maint).
 176
 177  * Traditionally we tried to avoid interpreting date strings given by
 178    the user as future dates, e.g. GIT_COMMITTER_DATE=2014-12-10 when
 179    used early November 2014 was taken as "October 12, 2014" because it
 180    is likely that a date in the future, December 10, is a mistake.
 181    This heuristics has been loosened to allow people to express future
 182    dates (most notably, --until=<date> may want to be far in the
 183    future) and we no longer tiebreak by future-ness of the date when
 184
 185     (1) ISO-like format is used, and
 186     (2) the string can make sense interpreted as both y-m-d and y-d-m.
 187
 188    Git may still have to use the heuristics to tiebreak between dd/mm/yy
 189    and mm/dd/yy, though.
 190    (merge d372395 jk/approxidate-avoid-y-d-m-over-future-dates later to maint).
 191
 192  * Git did not correctly read an overlong refname from a packed refs
 193    file.
 194    (merge ea41783 jk/read-packed-refs-without-path-max later to maint).
 195
 196  * "git apply" was described in the documentation to take --ignore-date
 197    option, which it does not.
 198    (merge 0cef4e7 rw/apply-does-not-take-ignore-date later to maint).
 199
 200  * "git add -i" did not notice when the interactive command input
 201    stream went away and kept asking the same question.
 202    (merge a8bec7a jk/add-i-read-error later to maint).
 203
 204  * "git send-email" did not handle RFC 2047 encoded headers quite
 205    right.
 206    (merge ab47e2a rd/send-email-2047-fix later to maint).
 207
 208  * New tag object format validation added in 2.2 showed garbage after
 209    a tagname it reported in its error message.
 210    (merge a1e920a js/fsck-tag-validation later to maint).
 211
 212  * The code that reads the reflog from the newer to the older entries
 213    did not handle an entry that crosses a boundary of block it uses to
 214    read them correctly.
 215    (merge 69216bf jk/for-each-reflog-ent-reverse later to maint).
 216
 217  * "git diff -B -M" after making a new copy B out of an existing file
 218    A and then editing A extensively ought to report that B was created
 219    by copying A and A was modified, which is what "git diff -C"
 220    reports, but it instead said A was renamed to B and A was edited
 221    heavily in place.  This was not just incoherent but also failed to
 222    apply with "git apply".  The report has been corrected to match what
 223    "git diff -C" produces for this case.
 224    (merge 6936b58 jc/diff-b-m later to maint).
 225
 226  * In files we pre-populate for the user to edit with commented hints,
 227    a line of hint that is indented with a tab used to show as '#' (or
 228    any comment char), ' ' (space), and then the hint text that began
 229    with the tab, which some editors flag as an indentation error (tab
 230    following space).  We now omit the space after the comment char in
 231    such a case.
 232    (merge d55aeb7 jc/strbuf-add-lines-avoid-sp-ht-sequence later to maint).
 233
 234  * "git ls-tree" does not support path selection based on negative
 235    pathspecs, but did not error out when negative pathspecs are given.
 236    (merge f1f6224 nd/ls-tree-pathspec later to maint).
 237
 238  * The function sometimes returned a non-freeable memory and some
 239    other times returned a piece of memory that must be freed, leading
 240    to inevitable leaks.
 241    (merge 59362e5 jc/exec-cmd-system-path-leak-fix later to maint).
 242
 243  * The code to abbreviate an object name to its short unique prefix
 244    has been optimized when no abbreviation was requested.
 245    (merge 61e704e mh/find-uniq-abbrev later to maint).
 246
 247  * "git add --ignore-errors ..." did not ignore an error to
 248    give a file that did not exist.
 249    (merge 1d31e5a mg/add-ignore-errors later to maint).
 250
 251  * "git checkout $treeish $path", when $path in the index and the
 252    working tree already matched what is in $treeish at the $path,
 253    still overwrote the $path unnecessarily.
 254    (merge c5326bd jk/checkout-from-tree later to maint).
 255
 256  * "git config --get-color" did not parse its command line arguments
 257    carefully.
 258    (merge cb35722 jk/colors-fix later to maint).
 259
 260  * open() emulated on Windows platforms did not give EISDIR upon
 261    an attempt to open a directory for writing.
 262    (merge ba6fad0 js/windows-open-eisdir-error later to maint).
 263
 264  * A few code paths used abs() when they should have used labs() on
 265    long integers.
 266    (merge 83915ba rs/maint-config-use-labs later to maint).
 267    (merge 31a8aa1 rs/receive-pack-use-labs later to maint).
 268
 269  * "gitweb" used to depend on a behaviour recent CGI.pm deprecated.
 270    (merge 13dbf46 jk/gitweb-with-newer-cgi-multi-param later to maint).
 271
 272  * "git init" (hence "git clone") initialized the per-repository
 273    configuration file .git/config with x-bit by mistake.
 274    (merge 1f32ecf mh/config-flip-xbit-back-after-checking later to maint).
 275
 276  * Recent update in Git 2.2 started creating objects/info/packs and
 277    info/refs files with permission bits tighter than user's umask.
 278    (merge d91175b jk/prune-packed-server-info later to maint).
 279
 280  * Git 2.0 was supposed to make the "simple" mode for the default of
 281    "git push", but it didn't.
 282    (merge 00a6fa0 jk/push-simple later to maint).
 283
 284  * "Everyday" document had a broken link.
 285    (merge 366c8d4 po/everyday-doc later to maint).
 286
 287  * A few test fixes.
 288    (merge 880ef58 jk/no-perl-tests later to maint).
 289
 290  * The build procedure did not bother fixing perl and python scripts
 291    when NO_PERL and NO_PYTHON build-time configuration changed.
 292    (merge ca2051d jk/rebuild-perl-scripts-with-no-perl-seting-change later to maint).
 293
 294  * The usage string of "git log" command was marked incorrectly for
 295    l10n.
 296    (merge e66dc0c km/log-usage-string-i18n later to maint).
 297
 298  * "git for-each-ref" mishandled --format="%(upstream:track)" when a
 299    branch is marked to have forked from a non-existing branch.
 300    (merge b6160d9 rc/for-each-ref-tracking later to maint).