Path to this page:
Next | Query returned 5 messages, browsing 1 to 10 | previous
CVS Commit History:
2014-08-29 16:09:59 by Ryosuke Moro | Files touched by this commit (104) |
Log message:
make it clear what package depend on
discussed with wiz@.
|
2013-11-09 00:07:20 by Ryosuke Moro | Files touched by this commit (2) |
Log message:
Update to 0.4.0.1
changes:
0.4.0.1
- Prevent parsing non-ordered containers
This comes from the discussion of this pull request:
https://github.com/GaloisInc/cereal/pull/20
It was pointed out that fromList on the containers will do the same check
internally that the pull request implements, so it seemed sensible to just
use that instead of replicate its functionality here.
|
2013-09-28 03:02:27 by Ryosuke Moro | Files touched by this commit (3) |
Log message:
Update to 0.4.0.0
changes:
0.4.0.0
- Update version number for release
* Major bump is due to new instances for types in Data.Monoid
- Merge pull request #18 from audreyt/patch-1
Unbreak cereal for GHC 7.6.3
- Merge pull request #14 from gcross/master
Instances for newtype wrappers in Data.Monoid
- Merge pull request #13 from alpmestan/master
Add -O2 to the ghc build options
- Fix a warning in ghc-7.4.1
- Add the remaining input to the Fail constructor
0.3.5.2
- Restructure runGetLazy to properly deal with the end of input
- Merge pull request #4 from meiersi/fix-remove-unnecessary-unpack
remove unnecessary UNPACK pragma
- Merge pull request #3 from meiersi/improve-encoding-performance
Improve encoding performance for lists, sequences, and trees.
- slight improvement 5% for Data.Tree encoding
- improved speed of Data.Sequence encoding
It's about a factor 1.5 faster. Thanks to using 'foldMap' instead of \
'mapM' it is slightly faster than the version in 'binary-0.5.0.2'.
- improve the speed of list encoding by up to a factor of 2
It is better to traverse the list twice than to explicitly construct a \
closure representing the encoding of all its elements, as this avoids a \
significant amout of allocation. We gain some additional speed by recursing over \
builders only instead of recursing in the 'Put' monad.
Another candidate for the first kind of optimization is 'putSeqOf' and one \
for the second kind of optimization is 'putTreeOf'.
- Merge pull request #2 from meiersi/fix-lazy-bytestring-length-conversion
make types explicit and avoid unnecessary roundtrip
- Warning removal
- Add check in getBytes for negative length
- Use the more efficient unsafeShiftR instead of shiftR when base >= 4.5
- Fixed compilation with recent base.
|
2012-03-04 08:59:16 by PHO / phonohawk | Files touched by this commit (4) | |
Log message:
Upstream update to cereal-0.3.5.1
|
2011-07-12 09:13:22 by PHO / phonohawk | Files touched by this commit (5) | |
Log message:
Import hs-cereal-0.3.3.0 as wip/hs-cereal.
A binary serialization library, similar to binary, that introduces an
isolate primitive for parser isolation, and replaces the asynchronous
errors with a user-handleable Either type. Similar to binary in
performance, but uses a strict ByteString instead of a lazy
ByteString, thus restricting it to operating on finite inputs.
|
Next | Query returned 5 messages, browsing 1 to 10 | previous