This is a bit of future-proofing esc_html and friends: when called
with undefined value they would now would return undef... which would
probably mean that error would still occur, but closer to the source
of problem.
This means that we can safely use
esc_html(shift) || "Internal Server Error"
in die_error() instead of
esc_html(shift || "Internal Server Error")
Signed-off-by: Jakub Narebski <jnareb@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The error message (second argument to die_error) is meant to be short,
one-line text description of given error. A few callers call
die_error with error message containing unescaped user supplied data
($hash, $file_name). Instead of forcing callers to escape data,
simply call esc_html on the parameter.
Note that optional third parameter, which contains detailed error
description, is meant to be HTML formatted, and therefore should be
not escaped.
While at it update esc_html synopsis/usage, and bring default error
description to read 'Internal Server Error' (titlecased).
Signed-off-by: Jakub Narebski <jnareb@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The chop_str subroutine is meant to be used on strings (such as commit
description / title) *before* HTML escaping, which means before
applying esc_html or equivalent.
Therefore get rid of the failed attempt to always remove full HTML
entities (like e.g. & or ). It is not necessary (HTML
entities gets added later), and it can cause chop_str to chop a string
incorrectly.
Specifically:
API & protocol: support option to force written data immediately to disk
from http://git.kernel.org/?p=daemon/distsrv/chunkd.git;a=commit;h=3b02f749df2cb1288f345a689d85e7061f507e54
The short version of the title gets chopped to
API ...
where it should be
API & protocol: support option to force written data...
Noticed-by: John 'Warthog9' Hawley <warthog9@kernel.org>
Signed-off-by: John 'Warthog9' Hawley <warthog9@kernel.org>
Signed-off-by: Jakub Narebski <jnareb@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Add a 3rd, optional, parameter to die_error to allow for extended error
information to be output along with what the error was.
Signed-off-by: John 'Warthog9' Hawley <warthog9@kernel.org>
Signed-off-by: Jakub Narebski <jnareb@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Add a function (named format_sort_th) that returns the string that
print_sort_th would print.
Signed-off-by: John 'Warthog9' Hawley <warthog9@kernel.org>
Signed-off-by: Jakub Narebski <jnareb@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Add a function (named format_local_time) that returns the string that
print_local_time would print.
Signed-off-by: John 'Warthog9' Hawley <warthog9@kernel.org>
Signed-off-by: Jakub Narebski <jnareb@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
If one of $site_header, $site_footer or $home_text is not defined, you
get extraneous errors in the web logs, for example (line wrapped for
better readibility):
[Wed Jan 13 16:55:42 2010] [error] [client ::1] [Wed Jan 13 16:55:42 2010]
gitweb.cgi: Use of uninitialized value $site_header in -f at
/var/www/gitweb/gitweb.cgi line 3287., referer: http://git/gitweb.cgi
This ensures that those variables are defined before trying to use it.
Note that such error can happen only because of an error in gitweb
config file; building gitweb.cgi can make mentioned variables holding
empty string (it is even the default), but they are still defined.
Signed-off-by: John 'Warthog9' Hawley <warthog9@kernel.org>
Signed-off-by: Jakub Narebski <jnareb@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Adjust the main Makefile so you can simply run
make gitweb
which in turn calls gitweb/Makefile. This means that in order to
generate gitweb, you can simply run 'make' from gitweb subdirectory:
cd gitweb
make
Targets gitweb/gitweb.cgi and (dependent on JSMIN being defined)
gitweb/gitweb.min.js in main Makefile are preserved for backward
compatibility.
Signed-off-by: John 'Warthog9' Hawley <warthog9@kernel.org>
Signed-off-by: Jakub Narebski <jnareb@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This changes slightly the behavior of gitweb, so that it verifies
that the box isn't inundated with before attempting to serve gitweb.
If the box is overloaded, it basically returns a 503 Server Unavailable
until the load falls below the defined threshold. This helps dramatically
if you have a box that's I/O bound, reaches a certain load and you
don't want gitweb, the I/O hog that it is, increasing the pain the
server is already undergoing.
This behavior is controlled by $maxload configuration variable.
Default is a load of 300, which for most cases should never be hit.
Unset it (set it to undefined value, i.e. undef) to turn off checking.
Currently it requires that '/proc/loadavg' file exists, otherwise the
load check is bypassed (load is taken to be 0). So platforms that do
not implement '/proc/loadavg' currently cannot use this feature
(provisions are included for additional checks to be added by others).
There is simple test in t/t9501-gitweb-standalone-http-status.sh to
check that it correctly returns "503 Service Unavailable" if load is
too high, and also if there are any Perl warnings or errors.
Signed-off-by: John 'Warthog9' Hawley <warthog9@kernel.org>
Signed-off-by: Jakub Narebski <jnareb@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
In Internet Explorer 8 (IE8) the 'blame_incremental' view, which uses
JavaScript to generate blame info using AJAX, sometimes hang at the
beginning (at 0%) of blaming, e.g. for larger files with long history
like git's own gitweb/gitweb.perl.
The error shown by JavaScript console is "Unspecified error" at char:2
of the following line in gitweb/gitweb.js:
if (xhr.readyState === 3 && xhr.status !== 200) {
Debugging it using IE8 JScript debuger shown that the error occurs
when trying to access xhr.status (xhr is XMLHttpRequest object).
Watch for xhr object shows 'Unspecified error.' as "value" of
xhr.status, and trying to access xhr.status from console throws error.
This bug is some intermittent bug, depending on XMLHttpRequest timing,
as it doesn't occur in all cases. It is probably caused by the fact
that handleResponse is called from timer (pollTimer), to work around
the fact that some browsers call onreadystatechange handler only once
for each state change, and not like required for 'blame_incremental'
as soon as new data is available from server. It looks like xhr
object is not properly initialized; still it is a bug to throw an
error when accessing xhr.status (and not use 'null' or 'undefined' as
value).
Work around this bug in IE8 by using try-catch block when accessing
xhr.status.
Signed-off-by: Jakub Narebski <jnareb@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
* jn/gitweb-blame:
gitweb: Add link to other blame implementation in blame views
gitweb: Make linking to actions requiring JavaScript a feature
gitweb.js: fix padLeftStr() and its usage
gitweb.js: Harden setting blamed commit info in incremental blame
gitweb.js: fix null object exception in initials calculation
gitweb: Minify gitweb.js if JSMIN is defined
gitweb: Create links leading to 'blame_incremental' using JavaScript
gitweb: Colorize 'blame_incremental' view during processing
gitweb: Incremental blame (using JavaScript)
gitweb: Add optional "time to generate page" info in footer
Conflicts:
Makefile
gitweb/gitweb.css
Add link to 'blame_incremental' action (which requires JavaScript) in
'blame' view, and add link to 'blame' action in 'blame_incremental'
view.
Signed-off-by: Jakub Narebski <jnareb@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Let gitweb turn some links (like 'blame' links) into linking to actions
which require JavaScript (like 'blame_incremental' action) only if
'javascript-actions' feature is enabled.
This means that links to such actions would be present only if both
JavaScript is enabled and 'javascript-actions' feature is enabled.
Signed-off-by: Jakub Narebski <jnareb@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
It seems that in Firefox-3.5 inserting with javascript inserts the
literal instead of a space. Fix this by inserting the unicode
representation for instead.
Also fix the off-by-one error in the padding calculation that was
causing one less space to be inserted than was requested by the caller.
Signed-off-by: Stephen Boyd <bebarino@gmail.com>
Cc: Jakub Narebski <jnareb@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Internet Explorer 8 stops at beginning of blame filling with the
following bug:
"firstChild is null or not an object"
at this line:
a_sha1.firstChild.data = commit.sha1.substr(0, 8);
It is (probably) caused by the fact that while a_sha1 element, which
looks like this:
<a href=""> </a>
It has a firstChild which is a text node containing only whitespace
(single space character) in other web browsers (Firefox 3.5, Opera 10,
Google Chrome 3.0), IE8 clobbers DOM, removing trailing/leading
whitespace.
Protect against this bug by creating text element if it does not
exist.
Found-by: Stephen Boyd <bebarino@gmail.com>
Signed-off-by: Jakub Narebski <jnareb@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
* mr/gitweb-snapshot:
t/gitweb-lib: Split HTTP response with non-GNU sed
gitweb: Smarter snapshot names
gitweb: Document current snapshot rules via new tests
t/gitweb-lib.sh: Split gitweb output into headers and body
gitweb: check given hash before trying to create snapshot
Currently handleLine() assumes that a commit author name will always
start with a capital letter. It's possible that the author name is
user@example.com and therefore calling a match() on the name will fail
to return any matches. Subsequently joining these matches will cause an
exception. Fix by checking that we have a match before trying to join
the results into a set of initials for the author.
Signed-off-by: Stephen Boyd <bebarino@gmail.com>
Acked-by: Jakub Narebski <jnareb@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Make git_history use git_log_generic, passing git_history_body as one
of its paramaters. This required changes to git_log_generic, in
particular passing more things as parameters.
While refactoring common code of 'log', 'shortlog' and 'history' view,
we did unify pagination, using always the form used by 'history' view,
namely
first * prev * next
in place of
HEAD * prev * next
used by 'log' and 'shortlog' views.
The 'history' view now supports commit limiting via 'hpb' parameter,
similarly to 'shortlog' (and 'log') view. Performance of 'history'
view got improved a bit, as it doesn't run git_get_hash_by_path for
"current" version in a loop. Error detection and reporting for
'history' view changed a bit.
Signed-off-by: Jakub Narebski <jnareb@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Put the common parts of git_log and git_shortlog into git_log_generic
subroutine: git_log and git_shortlog are now thin wrappers calling
git_log_generic with appropriate arguments.
The unification of code responsible for 'log' and 'shorlog' actions
lead to the following changes in gitweb output
* 'tree' link in page_nav now uses $hash parameter, as was the case
for 'shortlog' but not for 'log'
* 'log' view now respect $hash_parent limiting, like 'shortlog' did
* 'log' view doesn't have special case for empty list anymore, and it
always uses page_header linking to summary view, like 'shortlog'
did.
Signed-off-by: Jakub Narebski <jnareb@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Put the main part of 'log' view generation into git_log_body,
similarly how it is done for 'shortlog' and 'history' views (and
also for 'tags' and 'heads' views).
This is preparation for extracting common code between 'log',
'shortlog' and 'history' actions.
Signed-off-by: Jakub Narebski <jnareb@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Teach gitweb how to produce nicer snapshot names by only using the
short hash id. If clients make requests using a tree-ish that is not
a partial or full SHA-1 hash, then the short hash will also be appended
to whatever they asked for. If clients request snapshot of a tag
(which means that $hash ('h') parameter has 'refs/tags/' prefix),
use only tag name.
Update tests cases in t9502-gitweb-standalone-parse-output.
Gitweb uses the following format for snapshot filenames:
<sanitized project name>-<version info>.<snapshot suffix>
where <sanitized project name> is project name with '.git' or '/.git'
suffix stripped, unless '.git' is the whole project name. For
snapshot prefix it uses:
<sanitized project name>-<version info>/
as compared to <sanitized project name>/ before (without version info).
Current rules for <version info>:
* if 'h' / $hash parameter is SHA-1 or shortened SHA-1, use SHA-1
shortened to to 7 characters
* otherwise if 'h' / $hash parameter is tag name (it begins with
'refs/tags/' prefix, use tag name (with 'refs/tags/' stripped
* otherwise if 'h' / $hash parameter starts with 'refs/heads/' prefix,
strip this prefix, convert '/' into '.', and append shortened SHA-1
after '-', i.e. use <sanitized hash>-<shortened sha1>
Signed-off-by: Mark Rada <marada@uwaterloo.ca>
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Signed-off-by: Jakub Narebski <jnareb@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
In pathinfo mode, we use <base href> that refers to the base location
of gitweb in order for various external media links to work well.
However, this means that for the page to refer to itself, it must
regenerate full link, and this is exactly what the blob view page
did not do for line numbers.
Signed-off-by: Petr Baudis <pasky@suse.cz>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
It's nice to search for an author by merely clicking on their name in
gitweb. This is usually faster than selecting the name, copying the
selection, pasting it into the search box, selecting between
author/committer and then hitting enter.
Linkify the avatar icon in log/shortlog view because the icon is directly
adjacent to the name and thus more related. The same is not true
when in commit/tag view where the icon is farther away.
Signed-off-by: Stephen Boyd <bebarino@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The custom CGI escaping done in esc_param failed to escape UTF-8
properly. Fix by using CGI::escape on each sequence of matched
characters instead of sprintf()ing a custom escaping for each byte.
Additionally, the space -> + escape was being escaped due to greedy
matching on the first substitution. Fix by adding space to the
list of characters not handled on the first substitution.
Finally, remove an unnecessary escaping of the + sign.
Signed-off-by: Giuseppe Bilotta <giuseppe.bilotta@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The 'patch' view is about generating text/plain patch that can be
given to "git am", and "git am" doesn't understand merges anyway.
Therefore link to 'patch' view should not be shown for merge commits.
Also call to git-format-patch inside the 'patch' action would fail
when 'patch' action is called for a merge commit, with "Reading
git-format-patch failed" text as 'patch' view body.
Signed-off-by: Jakub Narebski <jnareb@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Makes things nicer in cases when you hand craft the snapshot URL but
make a typo in defining the hash variable (e.g. netx instead of next);
you will now get an error message instead of a broken tarball.
Tests for t9501 are included to demonstrate added functionality.
Signed-off-by: Mark Rada <marada@uwaterloo.ca>
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Add support for 'show-sizes' feature to show (in separate column,
between mode and filename) the size of blobs (files) in the 'tree'
view. It passes '-l' option to "git ls-tree" invocation.
For the 'tree' and 'commit' (submodule) entries, '-' is shown in place
of size; for generated '..' "up directory" entry nothing is shown.
The 'show-sizes' feature is enabled by default.
Signed-off-by: Jakub Narebski <jnareb@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The new 'blame_incremental' view requires JavaScript to run. Not all
web browsers implement JavaScript (e.g. text browsers such as Lynx),
and not all users have JavaScript enabled. Therefore instead of
unconditionally linking to 'blame_incremental' view, we use JavaScript
to convert those links to lead to view utilizing JavaScript, by adding
'js=1' to link.
Currently the only action that takes 'js=1' into account is 'blame',
which then acts as if it was called as 'blame_incremental' action.
Possible enhancement would be to do JavaScript redirect by setting
window.location instead of modifying $format and $action in
git_blame_common() subroutine.
The only JavaScript-aware/using view is currently 'blame_incremental'.
While at it move reading JavaScript to git_footer_html() subroutine.
Note that in this view we do not add 'js=1' currently (even though
perhaps we should; note that for consistency we should also add 'js=1'
in links added by JavaScript part of 'blame_incremental').
This idea was originally implemented by Petr Baudis in
http://article.gmane.org/gmane.comp.version-control.git/47614
but it added <script> element with fixBlameLinks() function in page
header, to be added as onload event using 'onload' attribute of HTML
'body' element: <body onload="fixBlameLinks();">. This version adds
script at then end of page (in the page footer), and uses JavaScript
'window.onload=fixLinks();'. Also in Petr version only links marked
with 'blamelink' class were modified, and they were modified by
replacing "a=blame" by "a=blame_incremental"... which doesn't work for
path_info links, and might replace wrong part if there is "a=blame" in
project name, ref name or file name.
Slightly different solution was implemented by Martin Koegler in
http://thread.gmane.org/gmane.comp.version-control.git/47902/focus=47905
Here GitAddLinks() function was in gitweb.js file, not as contents of
<script> element. It was also included in page header (in <head>
element) though, which means waiting for a script to load (and run).
It was smarter in that to "fix" (modify) link, it split URL, modified
value of 'a' parameter, and then recreated modified link. It avoids
trouble with "a=blame" as substring in project name or file name, but
it doesn't work with path_info URL/link in the way it was written.
Signed-off-by: Jakub Narebski <jnareb@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This requires using 3 colors, not only two, to choose a color that is
different from colors of up to 2 neighbors.
gitweb.js selects the least used color, if more than one color is
possible.
Signed-off-by: Jakub Narebski <jnareb@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Add 'blame_incremental' view, which uses "git blame --incremental"
and JavaScript (Ajax), where 'blame' use "git blame --porcelain".
* gitweb generates initial info by putting file contents (from
"git cat-file") together with line numbers in blame table
* then gitweb makes web browser JavaScript engine call startBlame()
function from gitweb.js
* startBlame() opens XMLHttpRequest connection to 'blame_data' view,
which in turn calls "git blame --incremental" for a file, and
streams output of git-blame to JavaScript (gitweb.js)
* XMLHttpRequest event handler updates line info in blame view as soon
as it gets data from 'blame_data' (from server), and it also updates
progress info
* when 'blame_data' ends, and gitweb.js finishes updating line info,
it fixes colors to match (as far as possible) ordinary 'blame' view,
and updates information about how long it took to generate page.
Gitweb deals with streamed 'blame_data' server errors by displaying
them in the progress info area (just in case).
The 'blame_incremental' view tries to be equivalent to 'blame' action;
there are however a few differences in output between 'blame' and
'blame_incremental' view:
* 'blame_incremental' always used query form for this part of link(s)
which is generated by JavaScript code. The difference is visible
if we use path_info link (pass some or all arguments in path_info).
Changing this would require implementing something akin to href()
subroutine from gitweb.perl in JavaScript (in gitweb.js).
* 'blame_incremental' always uses "rowspan" attribute, even if
rowspan="1". This simplifies code, and is not visible to user.
* The progress bar and progress info are still there even after
JavaScript part of 'blame_incremental' finishes work.
Note that currently no link generated by gitweb leads to this new view.
This code is based on patch by Petr Baudis <pasky@suse.cz> patch, which
in turn was tweaked up version of Fredrik Kuivinen <frekui@gmail.com>'s
proof of concept patch.
This patch adds GITWEB_JS compile configuration option, and modifies
git-instaweb.sh to take gitweb.js into account. The code for
git-instaweb.sh was taken from Pasky's patch.
Signed-off-by: Fredrik Kuivinen <frekui@gmail.com>
Signed-off-by: Petr Baudis <pasky@suse.cz>
Signed-off-by: Jakub Narebski <jnareb@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Add "This page took XXX seconds and Y git commands to generate"
to page footer, if global feature 'timed' is enabled (disabled
by default). Requires Time::HiRes installed for high precision
'wallclock' time.
Note that Time::HiRes is being required unconditionally; this is
because setting $t0 variable needs to be done fairly early to have
running time of the whole script. If Time::HiRes module were required
only if 'timed' feature is enabled, the earliest place where starting
time ($t0) could be calculated would be after reading gitweb config,
making "time to generate page" info inaccurate.
This code is based on example code by Petr 'Pasky' Baudis.
Signed-off-by: Jakub Narebski <jnareb@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
* mr/gitweb-snapshot:
gitweb: add t9501 tests for checking HTTP status codes
gitweb: split test suite into library and tests
gitweb: improve snapshot error handling
* maint:
Fix overridable written with an extra 'e'
Documentation: git-archive: mark --format as optional in summary
Round-down years in "years+months" relative date view
* maint-1.6.3:
Fix overridable written with an extra 'e'
Documentation: git-archive: mark --format as optional in summary
Round-down years in "years+months" relative date view
* maint-1.6.2:
Fix overridable written with an extra 'e'
Documentation: git-archive: mark --format as optional in summary
Round-down years in "years+months" relative date view
Conflicts:
Documentation/git-archive.txt
The last check in the second block of checks in the &git_snapshot routine
is never executed because the second to last check is a superset of the
last check.
Switch the order of the last two checks. It has the advantage of giving
clients a more specific reason why they cannot get a snapshot format if
the format they have chosen is disabled.
Signed-off-by: Mark Rada <marada@uwaterloo.ca>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
* mr/gitweb-xz:
gitweb: add support for XZ compressed snapshots
gitweb: update INSTALL regarding specific snapshot settings
gitweb: support to globally disable a snapshot format
Since 4afbaef (gitweb: ref markers link to named shortlogs, 2008-09-02),
ref markers that accompany the subject in views such as shortlog and
history point to something different from the subject itself. Therefore,
they should not be included in the same <a> element.
Benefits of the change are:
* better compliance to the XHTML standards, that forbid links within
links even though the restriction cannot be imposed via DTD; this also
benefits visualization in some older browsers;
* when hovering the subject, only the subject itself is underlined; when
hovering the ref markers, only the text in the hovered ref marker is
underlined; previously, hovering any written part of the subject column
led to complete underlying of everything at the same time, with
unpleasing effects.
Signed-off-by: Giuseppe Bilotta <giuseppe.bilotta@gmail.com>
Acked-by: Jakub Narebski <jnareb@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Reduce size of git-favicon.png using a combination of optipng and
pngout. From 164 bytes to 115 bytes (30% reduction). Also reduce
git-logo.png's size by one byte using advcomp.
Signed-off-by: Benjamin Kramer <benny.kra@googlemail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
* 'jn/gitweb-blame' (early part):
gitweb: Use light/dark for class names also in 'blame' view
gitweb: Add author initials in 'blame' view, a la "git gui blame"
gitweb: Mark commits with no "previous" in 'blame' view
gitweb: Use "previous" header of git-blame -p in 'blame' view
gitweb: Mark boundary commits in 'blame' view
gitweb: Make .error style generic
The XZ compression format uses the LZMA2 compression algorithm, which
often yields higher compression ratios than both GZip and BZip2 at the
cost of using more CPU time and RAM. XZ is the slowest for compression,
but still much faster than BZip2 for decompression, almost comparable
to GZip (see benchmarks below).
Some simple benchmarks show the pros and cons of using XZ compression;
starting with an already tarball'd archive of the repos listed below.
Memory usage seemed to be consistent for any given algorithm at their
respective default compression levels.
CPU: AMD Sempron 3400+ (1 core @ 1.8GHz with 256K L2 cache)
Virtual Memory Usage
GZip: 4152K BZip2: 13352K XZ: 102M
Linux 2.6 series (f5886c7f96f2542382d3a983c5f13e03d7fc5259) 349M
gzip 23.70s user 0.47s system 99% cpu 24.227 total 76M
gunzip 3.74s user 0.74s system 94% cpu 4.741 total
bzip2 130.96s user 0.53s system 99% cpu 2:11.97 total 59M
bunzip2 31.05s user 1.02s system 99% cpu 32.355 total
xz 448.78s user 0.91s system 99% cpu 7:31.28 total 51M
unxz 7.67s user 0.80s system 98% cpu 8.607 total
Git (0a53e9ddea) 11M
gzip 0.77s user 0.03s system 99% cpu 0.792 total 2.5M
gunzip 0.12s user 0.02s system 98% cpu 0.142 total
bzip2 3.42s user 0.02s system 99% cpu 3.454 total 2.1M
bunzip2 0.95s user 0.03s system 99% cpu 0.984 total
xz 12.88s user 0.14s system 98% cpu 13.239 total 1.9M
unxz 0.27s user 0.03s system 99% cpu 0.298 total
XZ (669413bb2db954bbfde3c4542fddbbab53891eb4) 1.8M
gzip 0.12s user 0.00s system 95% cpu 0.132 total 442K
gunzip 0.02s user 0.00s system 97% cpu 0.027 total
bzip2 1.28s user 0.01s system 99% cpu 1.298 total 363K
bunzip2 0.15s user 0.01s system 100% cpu 0.157 total
xz 1.62s user 0.03s system 99% cpu 1.652 total 347K
unxz 0.05s user 0.00s system 99% cpu 0.058 total
From a time and memory perspective, nothing compares to GZip, but if
given an average upload speed of 20KB/s, it would take ~400 seconds
longer to transfer the BZip2'd kernel snapshot than the XZ snapshot;
the transfer time difference is even greater between GZip and XZ. The
real time savings are relatively the same for all test cases, but less
dramatic for smaller repositories.
XZ decompresses ~1.8-2 times slower than GZip, and ~2.7-3.75 times
faster than BZip2; XZ gets relatively faster as snapshots get larger.
However, XZ takes relatively longer to compress as snapshots get larger.
The downside for XZ'd snapshots is the large CPU and memory load put on
the server to generate the compressed snapshot, though XZ will
eventually
have threading support, and the real clock time for making XZ'd
snapshots
would decrease if the server had a beefy multi-core CPU.
XZ compression is disabled by default to allow upgrades to take place
without any surprises, as the CPU and memory requirements will be an
issue for high load or lightweight servers. Also, the XZ format is still
new (format declared stable ~6 months ago), and there have been no
"stable" releases of the utils yet.
Signed-off-by: Mark Rada <marada@uwaterloo.ca>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This includes instructions on how to disable a snapshot format and how
to add options to a snapshot format (e.g. setting the compression
level).
Signed-off-by: Mark Rada <marada@uwaterloo.ca>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
