lp:~lifeless/meliae/stuff

Branch merges

Propose for merging

No branches dependent on this one.

Related source package recipes

No recipes using this branch. (?)

Create packaging recipe

Related snap packages

Related bugs

Related blueprints

86. By Robert Collins on 2009-09-13

Add some flush calls to reduce things being left in memory.

85. By John A Meinel on 2009-09-11

It didn't really seem safe to open the file in raw mode
to pass it into a Gzip stream.

84. By John A Meinel on 2009-09-11

Popen.terminate is a py2.6 thing.

So do the best workaround I could come up with.

83. By John A Meinel on 2009-09-11

In cygwin, gunzip is a shell script thunking over to gzip -d, while gzip is a real executable.
So just use 'gzip -d' directly. It is more likely to be available, anyway.

82. By John A Meinel on 2009-09-11

multiprocessing.Process only works with globals on Windows.

It uses cPickle to determine what function to run (since we don't have fork())
as such, the function to run must be globally accessible, and not a nested
function.

81. By John A Meinel on 2009-09-11

Cleanups to get the test suite passing with py25 and Windows.

80. By John A Meinel on 2009-09-11

Merge Robert's streaming work, will require some cleanups.

79. By John A Meinel on 2009-09-11

Declare a few functions static and inline. Minor benefit.

78. By John A Meinel on 2009-09-11

Do the same mutation to PyUnicode and _dump_object_to_ref_info

We now have a _write_static, which can just use strlen and doesn't
need to do any snprintf magic.
We are down to 6.6s.

77. By John A Meinel on 2009-09-11

Switching to a 'callable' meant that we doubled the time to dump 150MB.
However, changing the inner 'dump_c_string' function to work on a local buffer
allows us to do a lot of simple memory copying without any printf formatting.
Net result is that 10s => 19s and now => 7.8s.