test/regress/CAE63F5C-b.test and test/regress/CAE63F5C-c.test should both
pass, but test/regress/CAE63F5C-c.test does not, because the total line of
$6.46 is rounded wrong; it should be $6.45.
There seems to be different rounding occurring for totals vs. postings.
This seems to be related to Bug #492.
Due to weirdness that's currently true with the existing next branch of
ledger, I believe it's important to tell users in the documentation that
there are some discrepancies in the 'fixed' directive behavior.
The documentation from my previous commit is written to explain what
'fixed' *should* do; adding the bug report link here is a placeholder to
tell users that it may not do what they think it does.
Obviously, if someone closes#789, they should remove this paragraph added
herein. But, if the bug report is closed, but the documentation lags
behind, the worst that happens is some users have to click through to see
the bug is closed.
doc/ledger3.info was probably missing from the .gitignore because
ledger3.info isn't build automatically yet, but might as well add it to
.gitignore for those who are building it by hand at the moment.
Change the column of Receipt and Invoice in the CSV file first, then the
generated ODS file must have the same change propagated, which requires
changes to the column numbers hard-coding in csv2ods.py.
Perhaps if/when this application is refactored these things shouldn't be
hard-coded in this way in the first place.
I developed this, and therefore have the full git commit history, in my
personal "Small-Hacks" repository, which can be cloned from:
git://gitorious.org/bkuhn/small-hacks.git
More details on that are available by visiting:
https://gitorious.org/bkuhn/small-hacks
The basic idea here is that given non-profit-test-data.ledger herein,
there should be a script that I could run, in this fashion:
$ general-ledger-report -b 2011/03/01 -e 2012/03/01 -f tests/non-profit-test-data.ledger
that would generate:
non-profit-test-data_chart-of-accounts.txt
non-profit-test-data_general-ledger.ods
Note that the ODS file currently has placeholders, as I haven't fully
figured out how to use the =hyperlink() function to make relative
hyperlinks.
Upon discussion with John Wiegley <johnw@newartisans.com> on #ledger on
irc.freenode.net, the following was indicated:
<johnw> bkuhn: as long as the GPL infection stays in contrib, I see no problem
with it
...
<bkuhn> ... I got the ... answer, which is "johnw will accept GPL'd stuff
in contrib/..., as long as it's careful to not cause GPL to cover
the main Ledger codebase that's not in contrib/..."
Therefore, the non-profit-audit-reports/ application will be licensed
GPLv3-or-later.
Debian squeeze, which is currently the stable distribution at the time of
this commit, has both a Boost and a CMake that is too old for Ledger.
This FAQ entry explains how to build your own Boost and CMake for use with
Ledger, and the exact commands to type to build and install each, and then
configure, build and install Ledger against those new versions.
CMAKE_PREFIX_PATH is for searching for other programs, not for the place
to install this one. Based on acprep's --help, I think the intention was
to use CMAKE_INSTALL_PREFIX here.
Debian squeeze, which is currently the stable distribution at the time of
this commit, has both a Boost and a CMake that is too old for Ledger.
This FAQ entry explains how to build your own Boost and CMake for use with
Ledger, and the exact commands to type to build and install each, and then
configure, build and install Ledger against those new versions.
CMAKE_PREFIX_PATH is for searching for other programs, not for the place
to install this one. Based on acprep's --help, I think the intention was
to use CMAKE_INSTALL_PREFIX here.
Indeed, at the moment, it doesn't work against Python 3.x at all, so
ideally, we'd like to tell CMake that no Python versions except 2.7 and
2.6 are acceptable. However, at least as of CMake 2.8.8, there appears to
be no way to instruct CMake to never consider other versions of Python.
In other words, Python_ADDITIONAL_VERSIONS is prepended to the list of
possible Python versions considered, rather than replacing it wholly.
Theoretically, we could try to diddle withe the internal CMake variables
_PYTHON_FIND_OTHER_VERSIONS or _Python_VERSIONS somehow, but that seems
kludgey and dangerous. This patch is probably "enough for now" to at
least make sure that if the user has both Python 2.x and Python 3.x
installed, some version of 2.x that is known to work will be preferred.
