Commit by Richard Hopkins <richhguard-monotone@yahoo.co.uk> on net.venge.monotone.not-selector :: rcbbe6a548dd5d565dfd0601ba1df320a613d3dc6 / (2 files in 2 dirs): (link)

Implement 'not' selector function and corresponding tests

Example:
Show all revisions which have not been tested yet
mtn log -r not(c:testresult)

Commit by Richard Hopkins <richhguard-monotone@yahoo.co.uk> on net.venge.monotone :: r377d75d780a12a552941539f2e6a6e8d609fadb7 /src/cmd_diff_log.cc: (link)

Print old + new revision IDs in diff header when both are explicitly specified

Commit by Richard Hopkins <richhguard-monotone@yahoo.co.uk> on net.venge.monotone :: r1d7f0b547697172f9806535e1479911aed3ad783 /contrib/colortone.pl: (link)

Use 'require' + 'import' instead of 'use' for Win32::Console::ANSI

'use' results in a compile error if the module isn't found, such as on
non-windows systems.

Commit by Richard Hopkins <richhguard-monotone@yahoo.co.uk> on net.venge.monotone :: r14cf20bd261593dbe98b0d5c64bc5690892f03f2 /contrib/colortone.pl: (link)

Add colortone.pl which colors monotone output using ANSI escape codes

Windows users should probably create the '.bat' file as mentioned in the
file comments; makes using it much easier.

Commit by Richard Hopkins <richhguard-monotone@yahoo.co.uk> on net.venge.monotone :: rd69b6671da387f29b2acafba5ade90aaa190d0da / (3 files in 2 dirs): (link)

Produce meaningful error message when trying to disapprove a root

The old code used to generate an invariant failure in get_revision, called
from get_revision_manifest. This is because root revisions have no parents,
making the code trying to get the null_id revision.

Rather than let this happen, produce a friendlier message but still error
out as we don't handle disapproving of root revisions.

Commit by Richard Hopkins <richhguard-monotone@yahoo.co.uk> on net.venge.monotone :: rb95f29a31b0bdfcefe0a168dfa6d8b125ff96e1e /src/ (work.hh pcrewrap.cc): (link)

merge of 'a9d1085eb0df7cc8a49d22fbf80d4209cbca564d'
     and 'ef60d096fde0204e1884af172d2b901513a106f0'

Commit by Richard Hopkins <richhguard-monotone@yahoo.co.uk> on net.venge.monotone :: ra9d1085eb0df7cc8a49d22fbf80d4209cbca564d /src/pcrewrap.cc: (link)

Use const_iterator instead of iterator for regex_cache

find() and end() are only used to see if an entry exists in the cache, and
have no need to modify the results.

Commit by Richard Hopkins <richhguard-monotone@yahoo.co.uk> on net.venge.monotone :: ref60d096fde0204e1884af172d2b901513a106f0 /src/work.hh: (link)

Use 'struct i18n_format' instead of 'class i18n_format' in 'work.hh'

'sanity.hh' defines 'i18n_format' as a struct, and all other uses are
structs as well.

Commit by Richard Hopkins <richhguard-monotone@yahoo.co.uk> on net.venge.monotone.regex-cache :: r8ff9be8a9b6534d3c362b61258d724809d1e62e7 /NEWS: (link)

Add notice to NEWS file about performance of regular expressions

Commit by Richard Hopkins <richhguard-monotone@yahoo.co.uk> on net.venge.monotone.regex-cache :: rdbb5af793bce29645edde73368225c5259a06ecc /src/pcrewrap.cc: (link)

Create 'regex_cache_manager' to manage the regex cache and correctly free
the compiled data

Commit by Richard Hopkins <richhguard-monotone@yahoo.co.uk> on net.venge.monotone.regex-cache :: r3f8f0d347ad197aafa02785464810636e5db07eb /src/pcrewrap.cc: (link)

Cache previously compiled regexs in pcrewrap.cc

The benefit of this is to improve the use of the ".mtn-ignore" file by
commands such as "mtn ls unknown".

The timings obtained via "time mtn ls unknown" are shown below when
run on a Dell Inspiron 1501 with 894 MB RAM, OpenSUSE 11.4. The workspace
was a normal checkout of "net.venge.monotone" after the standard
"./configure", "make" cycle and using the standard ".mtn-ignore" file as
part of the checkout.

Repeated execution obtains results consistent with those below showing the
cache improves execution time.

Normal monotone (no caching)

real: 0m0.365s
user: 0m0.312s
sys: 0m0.052s

With caching

real: 0m0.316s
user: 0m0.240s
sys: 0m0.072s

However, currently the cache will be leaked as I don't know where to call
the new function "free_compiled" which will free everything.

Commit by Richard Levitte <richard@levitte.org> on net.venge.monotone :: rc954e1afd2121eff60933e93270d855a81d726ad /Makefile.am: (link)

• Makefile.am: Correct typo, $(test_unit_tester_SOURCES) should be $(test_bin_unit_tester_SOURCES). Closes issue 195

Commit by Richard Levitte <richard@levitte.org> on net.venge.monotone :: raf378ab04141f724f5c49d599e96b0f61c3cf2df /doc/pcrepattern.texi: (link)

• doc/pcrepattern.texi: Correct typo, "@verbatim" should be ended with "@end verbatim". Closes issue 196.

Commit by Richard Levitte <richard@levitte.org> on net.venge.monotone :: rb364caeea5e26e19023b68a166f14ee4127996e5 /Attic/pcre/README: (link)

• Attic/pcre/README: Update version number to match pcre.h. Closes issue 197.

Commit by Richard Levitte <richard@levitte.org> on net.venge.monotone :: re3bb68c53c351f411ea4505400c16c8d9f252013 /src/cmd.cc: (link)

• src/cmd.cc: Since the first string isn't a format string, let's not pretend it is. Instead, use fputs(), that's more efficient. The backgrond for this is Debian, where there are options to build with the following settings: CFLAGS=-g -O2 -fstack-protector --param=ssp-buffer-size=4 -Wformat -Wformat-security -Werror=format-security CPPFLAGS=-D_FORTIFY_SOURCE=2 CXXFLAGS=-g -O2 -fstack-protector --param=ssp-buffer-size=4 -Wformat -Wformat-security -Werror=format-security

Commit by Richard Hopkins <richhguard-monotone@yahoo.co.uk> on net.venge.monotone.log-with-certs :: r485dbe5582f16e43a91d56590d402808d4837990 /src/rev_output.cc: (link)

Output all custom certs under "Other certs" heading

NOTE: "suspend" and "testresult" certs, which are special to monotone, will
currently appear in the "Other certs" section. These will later get
promoted to their own section with their own special formatting.

Commit by Martin.vGagern@gmx.net on net.venge.monotone.issue-182 :: r0567ddefc130f03116a1b8a4e3a9a277a9da87af /src/monotone.cc: (link)

Adjust botan runtime version check to accept 1.10.x.

Commit by Martin.vGagern@gmx.net on net.venge.monotone.issue-182 :: r754a7a4e12aafce0cf45ab00329f0b6513502066 /src/ (8 files): (link)

Make monotone compile against botan 1.10.1.

Commit by Martin.vGagern@gmx.net on net.venge.monotone.issue-182 :: r30c022e6012f74f2b407b0cba7782fce17a041f4 /src/ (key_store.cc packet.cc packet.hh): (link)

Detect encoded keys with botan 1.10.x.

Older botan releases used to throw an Invalid_Argument exception in case of
an empty pass phrase, which in turn was derived from Botan::Exception.
Neither is necessarily true now, mostly due to botan commits
2d09d7d0cd4bd0e7155d001dd65a4f29103b158c and
e30a741f6014987e9683842ede22c44b2db03c56. So now we use our own dummy UI to
throw a well-known exception if a passphrase would be required.

At least for key_store, actually doing the user interaction inside a
User_Interface specialization would probably be the better solution in the
long run, but this dirty hack does the job for now.

Commit by Martin.vGagern@gmx.net on net.venge.monotone.issue-182 :: rdc0bcdde415716d2a6cd751b2fb5315d0daec265 /src/transforms.cc: (link)

Make transform failures recoverable even with botan 1.10.x.

We have to deal with the fact that Botan::Invalid_Argument no longer derives
from Botan::Exception. So now we use std::exception as the common base
class which we handle in error_in_transform.