Documentation/RelNotes/2.1.0.txt

   1 Git v2.1 Release Notes
   2 ======================
   3
   4 Backward compatibility notes
   5 ----------------------------
   6
   7  * The default value we give to the environment variable LESS has been
   8    changed from "FRSX" to "FRX", losing "S" (chop long lines instead
   9    of wrapping).  Existing users who prefer not to see line-wrapped
  10    output may want to set
  11
  12      $ git config core.pager "less -S"
  13
  14    to restore the traditional behaviour.  It is expected that people
  15    find output from most subcommands easier to read with the new
  16    default, except for "blame" which tends to produce really long
  17    lines.  To override the new default only for "git blame", you can
  18    do this:
  19
  20      $ git config pager.blame "less -S"
  21
  22  * A few disused directories in contrib/ have been retired.
  23
  24
  25 Updates since v2.0
  26 ------------------
  27
  28 UI, Workflows & Features
  29
  30  * Since the very beginning of Git, we gave the LESS environment a
  31    default value "FRSX" when we spawn "less" as the pager.  "S" (chop
  32    long lines instead of wrapping) has been removed from this default
  33    set of options, because it is more or less a personal taste thing,
  34    as opposed to the others that have good justifications (i.e. "R" is
  35    very much justified because many kinds of output we produce are
  36    colored and "FX" is justified because output we produce is often
  37    shorter than a page).
  38
  39  * The logic and data used to compute the display width needed for
  40    UTF-8 strings have been updated to match Unicode 7.0 better.
  41
  42  * HTTP-based transports learned to better propagate the error messages from
  43    the webserver to the client coming over the HTTP transport.
  44
  45  * The completion script for bash (in contrib/) has been updated to
  46    better handle aliases that define a complex sequence of commands.
  47
  48  * The "core.preloadindex" configuration variable is enabled by default,
  49    allowing modern platforms to take advantage of their
  50    multiple cores.
  51
  52  * "git clone" applies the "if cloning from a local disk, physically
  53    copy the repository using hardlinks, unless otherwise told not to with
  54    --no-local" optimization when the url.*.insteadOf mechanism rewrites a
  55    remote-repository "git clone $URL" into a
  56    clone from a local disk.
  57
  58  * "git commit --date=<date>" option learned more
  59    timestamp formats, including "--date=now".
  60
  61  * The `core.commentChar` configuration variable is used to specify a
  62    custom comment character (other than the default "#") for
  63    the commit message editor.  This can be set to `auto` to attempt to
  64    choose a different character that does not conflict with any that
  65    already starts a line in the message being edited, for cases like
  66    "git commit --amend".
  67
  68  * "git format-patch" learned --signature-file=<file> to add the contents
  69    of a file as a signature to the mail message it produces.
  70
  71  * "git grep" learned the grep.fullname configuration variable to force
  72    "--full-name" to be the default.  This may cause regressions for
  73    scripted users who do not expect this new behaviour.
  74
  75  * "git imap-send" learned to ask the credential helper for auth
  76    material.
  77
  78  * "git log" and friends now understand the value "auto" for the
  79    "log.decorate" configuration variable to enable the "--decorate"
  80    option automatically when the output is sent to tty.
  81
  82  * "git merge" without an argument, even when there is an upstream
  83    defined for the current branch, refused to run until
  84    merge.defaultToUpstream is set to true.  Flip the default of that
  85    configuration variable to true.
  86
  87  * "git mergetool" learned to drive the vimdiff3 backend.
  88
  89  * mergetool.prompt used to default to 'true', always asking "do you
  90    really want to run the tool on this path?".  The default has been
  91    changed to 'false'.  However, the prompt will still appear if
  92    mergetool used its autodetection system to guess which tool to use.
  93    Users who explicitly specify or configure a tool will no longer see
  94    the prompt by default.
  95
  96    Strictly speaking, this is a backward incompatible change and
  97    users need to explicitly set the variable to 'true' if they want
  98    to be prompted to confirm running the tool on each path.
  99
 100  * "git replace" learned the "--edit" subcommand to create a
 101    replacement by editing an existing object.
 102
 103  * "git replace" learned a "--graft" option to rewrite the parents of a
 104    commit.
 105
 106  * "git send-email" learned "--to-cover" and "--cc-cover" options, to
 107    tell it to copy To: and Cc: headers found in the first input file
 108    when emitting later input files.
 109
 110  * "git svn" learned to cope with malformed timestamps with only one
 111    digit in the hour part, e.g. 2014-01-07T5:01:02.048176Z, emitted
 112    by some broken subversion server implementations.
 113
 114  * "git tag" when editing the tag message shows the name of the tag
 115    being edited as a comment in the editor.
 116
 117  * "git tag" learned to pay attention to "tag.sort" configuration, to
 118    be used as the default sort order when no --sort=<value> option
 119    is given.
 120
 121  * A new "git verify-commit" command, to check GPG signatures in signed
 122    commits, in a way similar to "git verify-tag" is used to check
 123    signed tags, was added.
 124
 125
 126 Performance, Internal Implementation, etc.
 127
 128  * Build procedure for 'subtree' (in contrib/) has been cleaned up.
 129
 130  * Support for the profile-feedback build, which has
 131    bit-rotted for quite a while, has been updated.
 132
 133  * An experimental format to use two files (the base file and
 134    incremental changes relative to it) to represent the index has been
 135    introduced; this may reduce I/O cost of rewriting a large index
 136    when only small part of the working tree changes.
 137
 138  * Effort to shrink the size of patches Windows folks maintain on top
 139    by upstreaming them continues.  More tests that are not applicable
 140    to the Windows environment are identified and either skipped or
 141    made more portable.
 142
 143  * Eradication of "test $condition -a $condition" from our scripts
 144    continues.
 145
 146  * The `core.deltabasecachelimit` used to default to 16 MiB , but this
 147    proved to be too small, and has been bumped to 96 MiB.
 148
 149  * "git blame" has been optimized greatly by reorganising the data
 150    structure that is used to keep track of the work to be done.
 151
 152  * "git diff" that compares 3-or-more trees (e.g. parents and the
 153    result of a merge) has been optimized.
 154
 155  * The API to update/delete references are being converted to handle
 156    updates to multiple references in a transactional way.  As an
 157    example, "update-ref --stdin [-z]" has been updated to use this
 158    API.
 159
 160  * skip_prefix() and strip_suffix() API functions are used a lot more
 161    widely throughout the codebase now.
 162
 163  * Parts of the test scripts can be skipped by using a range notation,
 164    e.g. "sh t1234-test.sh --run='1-4 6 8-'" to omit test piece 5 and 7
 165    and run everything else.
 166
 167
 168 Also contains various documentation updates and code clean-ups.
 169
 170
 171 Fixes since v2.0
 172 ----------------
 173
 174 Unless otherwise noted, all the fixes since v2.0 in the maintenance
 175 track are contained in this release (see the maintenance releases'
 176 notes for details).
 177
 178  * We used to unconditionally disable the pager in the pager process
 179    we spawn to feed out output, but that prevented people who want to
 180    run "less" within "less" from doing so.
 181    (merge c0459ca je/pager-do-not-recurse later to maint).
 182
 183  * Tools that read diagnostic output in our standard error stream do
 184    not want to see terminal control sequence (e.g. erase-to-eol).
 185    Detect them by checking if the standard error stream is connected
 186    to a tty.
 187    (merge 38de156 mn/sideband-no-ansi later to maint).
 188
 189  * Mishandling of patterns in .gitignore that have trailing SPs quoted
 190    with backslashes (e.g. ones that end with "\ ") has been
 191    corrected.
 192    (merge 97c1364be6b pb/trim-trailing-spaces later to maint).
 193
 194  * Reworded the error message given upon a failure to open an existing
 195    loose object file due to e.g. permission issues; it was reported as
 196    the object being corrupt, but that is not quite true.
 197    (merge d6c8a05 jk/report-fail-to-read-objects-better later to maint).
 198
 199  * "git log -2master" is a common typo that shows two commits starting
 200    from whichever random branch that is not 'master' that happens to
 201    be checked out currently.
 202    (merge e3fa568 jc/revision-dash-count-parsing later to maint).
 203
 204  * Code to avoid adding the same alternate object store twice was
 205    subtly broken for a long time, but nobody seems to have noticed.
 206    (merge 80b4785 rs/fix-alt-odb-path-comparison later to maint).
 207    (merge 539e750 ek/alt-odb-entry-fix later to maint).
 208
 209  * The "%<(10,trunc)%s" pretty format specifier in the log family of
 210    commands is used to truncate the string to a given length (e.g. 10
 211    in the example) with padding to column-align the output, but did
 212    not take into account that number of bytes and number of display
 213    columns are different.
 214    (merge 7d50987 as/pretty-truncate later to maint).
 215
 216  * "%G" (nothing after G) is an invalid pretty format specifier, but
 217    the parser did not notice it as garbage.
 218    (merge 958b2eb jk/pretty-G-format-fixes later to maint).
 219
 220  * A handful of code paths had to read the commit object more than
 221    once when showing header fields that are usually not parsed.  The
 222    internal data structure to keep track of the contents of the commit
 223    object has been updated to reduce the need for this double-reading,
 224    and to allow the caller find the length of the object.
 225    (merge 218aa3a jk/commit-buffer-length later to maint).
 226
 227  * The "mailmap.file" configuration option did not support tilde
 228    expansion (i.e. ~user/path and ~/path).
 229    (merge 9352fd5 ow/config-mailmap-pathname later to maint).
 230
 231  * The completion scripts (in contrib/) did not know about quite a few
 232    options that are common between "git merge" and "git pull", and a
 233    couple of options unique to "git merge".
 234    (merge 8fee872 jk/complete-merge-pull later to maint).
 235
 236  * The unix-domain socket used by the sample credential cache daemon
 237    tried to unlink an existing stale one at a wrong path, if the path
 238    to the socket was given as an overlong path that does not fit in
 239    the sun_path member of the sockaddr_un structure.
 240    (merge 2869b3e rs/fix-unlink-unix-socket later to maint).
 241
 242  * An ancient rewrite passed a wrong pointer to a curl library
 243    function in a rarely used code path.
 244    (merge 479eaa8 ah/fix-http-push later to maint).
 245
 246  * "--ignore-space-change" option of "git apply" ignored the spaces
 247    at the beginning of lines too aggressively, which is inconsistent
 248    with the option of the same name that "diff" and "git diff" have.
 249    (merge 14d3bb4 jc/apply-ignore-whitespace later to maint).
 250
 251  * "git blame" miscounted the number of columns needed to show localized
 252    timestamps, resulting in a jaggy left-side-edge for the source code
 253    lines in its output.
 254    (merge dd75553 jx/blame-align-relative-time later to maint).
 255
 256  * "git blame" assigned the blame to the copy in the working-tree if
 257    the repository is set to core.autocrlf=input and the file used CRLF
 258    line endings.
 259    (merge 4d4813a bc/blame-crlf-test later to maint).
 260
 261  * "git clone -b brefs/tags/bar" would have mistakenly thought we were
 262    following a single tag, even though it was a name of the branch,
 263    because it incorrectly used strstr().
 264    (merge 60a5f5f jc/fix-clone-single-starting-at-a-tag later to maint).
 265
 266  * "git commit --allow-empty-message -C $commit" did not work when the
 267    commit did not have any log message.
 268    (merge 076cbd6 jk/commit-C-pick-empty later to maint).
 269
 270  * "git diff --find-copies-harder" sometimes pretended as if the mode
 271    bits have changed for paths that are marked with the assume-unchanged
 272    bit.
 273    (merge 5304810 jk/diff-files-assume-unchanged later to maint).
 274
 275  * "filter-branch" left an empty single-parent commit that results when
 276    all parents of a merge commit get mapped to the same commit, even
 277    under "--prune-empty".
 278    (merge 79bc4ef cb/filter-branch-prune-empty-degenerate-merges later to maint).
 279
 280  * "git format-patch" did not enforce the rule that the "--follow"
 281    option from the log/diff family of commands must be used with
 282    exactly one pathspec.
 283    (merge dd63f16 jk/diff-follow-must-take-one-pathspec later to maint).
 284
 285  * "git gc --auto" was recently changed to run in the background to
 286    give control back early to the end-user sitting in front of the
 287    terminal, but it forgot that housekeeping involving reflogs should
 288    be done without other processes competing for accesses to the refs.
 289    (merge 62aad18 nd/daemonize-gc later to maint).
 290
 291  * "git grep -O" to show the lines that hit in the pager did not work
 292    well with case insensitive search.  We now spawn "less" with its
 293    "-I" option when it is used as the pager (which is the default).
 294    (merge f7febbe sk/spawn-less-case-insensitively-from-grep-O-i later to maint).
 295
 296  * We used to disable threaded "git index-pack" on platforms without
 297    thread-safe pread(); use a different workaround for such
 298    platforms to allow threaded "git index-pack".
 299    (merge 3953949 nd/index-pack-one-fd-per-thread later to maint).
 300
 301  * The error reporting from "git index-pack" has been improved to
 302    distinguish missing objects from type errors.
 303    (merge 77583e7 jk/index-pack-report-missing later to maint).
 304
 305  * "log --show-signature" incorrectly decided the color to paint a
 306    mergetag that was and was not correctly validated.
 307    (merge 42c55ce mg/fix-log-mergetag-color later to maint).
 308
 309  * "log --show-signature" did not pay attention to the "--graph" option.
 310    (merge cf3983d zk/log-graph-showsig later to maint).
 311
 312  * "git mailinfo" used to read beyond the ends of header strings while
 313    parsing an incoming e-mail message to extract the patch.
 314    (merge b1a013d rs/mailinfo-header-cmp later to maint).
 315
 316  * On a case insensitive filesystem, merge-recursive incorrectly
 317    deleted the file that is to be renamed to a name that is the same
 318    except for case differences.
 319    (merge baa37bf dt/merge-recursive-case-insensitive later to maint).
 320
 321  * Merging changes into a file that ends in an incomplete line made the
 322    last line into a complete one, even when the other branch did not
 323    change anything around the end of file.
 324    (merge ba31180 mk/merge-incomplete-files later to maint).
 325
 326  * "git pack-objects" unnecessarily copied the previous contents when
 327    extending the hashtable, even though it will populate the table
 328    from scratch anyway.
 329    (merge fb79947 rs/pack-objects-no-unnecessary-realloc later to maint).
 330
 331  * Recent updates to "git repack" started to duplicate objects that
 332    are in packfiles marked with the .keep flag into the new packfile by
 333    mistake.
 334    (merge d078d85 jk/repack-pack-keep-objects later to maint).
 335
 336  * "git rerere forget" did not work well when merge.conflictstyle
 337    was set to a non-default value.
 338    (merge de3d8bb fc/rerere-conflict-style later to maint).
 339
 340  * "git remote rm" and "git remote prune" can involve removing many
 341    refs at once, which is not a very efficient thing to do when very
 342    many refs exist in the packed-refs file.
 343    (merge e6bea66 jl/remote-rm-prune later to maint).
 344
 345  * "git log --exclude=<glob> --all | git shortlog" worked as expected,
 346    but "git shortlog --exclude=<glob> --all", which is supposed to be
 347    identical to the above pipeline, was not accepted at the command
 348    line argument parser level.
 349    (merge eb07774 jc/shortlog-ref-exclude later to maint).
 350
 351  * The autostash mode of "git rebase -i" did not restore the dirty
 352    working tree state if the user aborted the interactive rebase by
 353    emptying the insn sheet.
 354    (merge ddb5432 rr/rebase-autostash-fix later to maint).
 355
 356  * "git rebase --fork-point" did not filter out patch-identical
 357    commits correctly.
 358
 359  * During "git rebase --merge", a conflicted patch could not be
 360    skipped with "--skip" if the next one also conflicted.
 361    (merge 95104c7 bc/fix-rebase-merge-skip later to maint).
 362
 363  * "git show -s" (i.e. show log message only) used to incorrectly emit
 364    an extra blank line after a merge commit.
 365    (merge ad2f725 mk/show-s-no-extra-blank-line-for-merges later to maint).
 366
 367  * "git status", even though it is a read-only operation, tries to
 368    update the index with refreshed lstat(2) info to optimize future
 369    accesses to the working tree opportunistically, but this could
 370    race with a "read-write" operation that modifies the index while it
 371    is running.  Detect such a race and avoid overwriting the index.
 372    (merge 426ddee ym/fix-opportunistic-index-update-race later to maint).
 373
 374  * "git status" (and "git commit") behaved as if changes in a modified
 375    submodule are not there if submodule.*.ignore configuration is set,
 376    which was misleading.  The configuration is only to unclutter diff
 377    output during the course of development, and not to hide
 378    changes in the "status" output to cause the users forget to commit
 379    them.
 380    (merge c215d3d jl/status-added-submodule-is-never-ignored later to maint).
 381
 382  * Documentation for "git submodule sync" forgot to say that the subcommand
 383    can take the "--recursive" option.
 384    (merge 9393ae7 mc/doc-submodule-sync-recurse later to maint).
 385
 386  * "git update-index --cacheinfo" in 2.0 release crashed on a
 387    malformed command line.
 388    (merge c8e1ee4 jc/rev-parse-argh-dashed-multi-words later to maint).
 389
 390  * The mode to run tests with HTTP server tests disabled was broken.
 391    (merge afa53fe na/no-http-test-in-the-middle later to maint).