We had a fairly ugly violation of modularity with the top-level
Makefile.local isntalling everything, (even when the build commands
for the library were down in lib/Makefile.local).
One of the supproted mechanisms we offer for configuration is
manually editing the Makefile.config file after it is generated
by the configure script. In this case it would be nice to be able
to change prefix only once, so allow that.
There's not any special configure logic for determining these variable
values, but if we did add some in the future, then these will now be
in the right place for that.
Additionally, this now makes Makefile.local the single place for the
user to look for manually tweaking a variable assignment, (say, for a
compiler that can't accept a particular warning argument).
With this change, there should rarely be any need for a user to poke
into any Makefile.local file.
Before it was impossible to know whether any particular setting or
rule definition was in Makefile or Makefile.local. So we strip the
Makefile down to little more than the list of sub-directories and
the logic to include all of the sub-directories' Makefile.local
fragments.
Then, all of the real work can happen inside of Makefile.local.
If an explicit --libdir is passed, then that is used directly. Otherwise
libdir is chosen as the value of $PREFIX/lib, (whether or not prefix was
passed explicitly or set by default).
For the case of adding a file that already exist, (with the same
filename). In this case, nothing will happen to the database, but
that wasn't clear before.
When trying to restore the current position, if the "current" thread
no longer appears in the buffer, then '=' moves to the current line
instead. When near the end of the buffer, the "current" line should
be counted as the number of lines from the end.
I just tried (and failed) to write a test for the recent magic
inference of phrase searches. That's a feature that makes me *really*
uncomfortable to not have an automated test. But I believe the
proposed modularization of the test suite should reduce some quoting
nightmares, so will hopefully make this easier.
These results have all the same terms as the target phrase, but
not in the expected order. They are designed to ensure that we
actually test phrase searches.
And as it turns out, we're not currently quoting the search terms
properly, so the phrase-search tests now fail with this commit.
Sebastian:
This replaces the patch it responds to. With this patch, we can now use
the cnotmuch with David's json ui. There are still issues, but this
allows interaction with emacs.
---
We are still missing Database.upgrade() as I am not sure how to implement the callback, and it's not that important for now. Documentation for the new classes is written inline, but not integrated in the docs yet.
