aboutsummaryrefslogtreecommitdiff
path: root/src (follow)
AgeCommit message (Collapse)Author
2011-10-27Added simple hex encoder/decoder filter.Mike Buland
2011-10-26Virtual deconstructor!Mike Buland
2011-10-25Base64 does line wrapping correctly on write, and also doesn't try to flush theMike Buland
write buffer when reading is done. It's...strange, but yeah, it was doing that. Deflate also defaults to zlib compression now, which means you can compress & decompress without using any extra params. Turns out zlib auto-detect won't decompress raw streams, so this is the safest overall option, and the easiest to work with. zlib headers are small, and includes a crc at the end so you can be sure your data is accurate, raw does not.
2011-10-25Fixed bug in base64 decoding. If an attempt is made to read data after the endMike Buland
of the stream has been reached, and the input didn't end with '=' chars then it would return the final buffer an extra time before ending. Now it ends when it should, no matter how many extra times you try to read.
2011-10-25Bu::print and Bu::println are added, they just take a string, but you can useMike Buland
string formatting no problem. I'll add some that can take unicode strings too.
2011-10-24Added some more docs.Mike Buland
2011-10-24Made the encoder state opaque to the caller in Deflate and BZip2 to match Lzma.Mike Buland
That means that when you use Bu::Deflate, Bu::Bzip2, or Bu::Lzma you don't get any of the respective libraries' header files.
2011-10-24Lzma filter added. Now we're really getting somewhere. Libbu++ now supportsMike Buland
all major, common compression algorithms.
2011-10-23The strfilter.h header has been added, it gives you one call filter access onMike Buland
strings. encodeStr<Filter>( str ) and decodeStr<Filter>( str ). It's pretty cool, try it out :)
2011-10-23Switched ito* to synchro*, except the server, I'm thinking of takeing the coreMike Buland
in a different direction anyway. Added the Deflate class, it uses zlib, and can do raw (headerless) deflate streams, zlib format, or gzip format. It's easy to use and quite versitile.
2011-10-06Renamed most of the core threading system, some ancillary systems need someMike Buland
kind of prefix or something, we could stick with Ito, I will until I think of something else.
2011-10-06Fixed the formatting of unsigned chars in the formatter, and added a function toMike Buland
actually stop the fastcgi main loop.
2011-09-24Fixed a 32/64 bit bug in the hash tables.Mike Buland
2011-09-16More helper functions for Bu::Process.Mike Buland
2011-08-25Bu::ItoLocker is rather handy, I like it. I reccomend it.Mike Buland
2011-08-15You can setgid now too.Mike Buland
2011-08-15Process now supports setuid for the child process. Really cool.Mike Buland
2011-07-27More features for uuids, including hashing and comparison.Mike Buland
2011-07-25Some tweaks to the uuid class to make it easier to work with, next for it isMike Buland
probably going to be archiver operators.
2011-07-25In theory, after constructing a TcpSocket object, you can now take theMike Buland
operating system socket handle away and use it yourself, after doing so, the TcpSocket will not close the socket when it exits, or be able to effect it ever again.
2011-07-20Oh wow, floats didn't have signs. Scary.Mike Buland
2011-07-17More variant unit tests, signed and unsigned chars are no formatted as numbers,Mike Buland
while chars are formatted as characters.
2011-07-13Trying to hunt down bug in variant code.Mike Buland
2011-06-22Signals.h is now autogenerated by build. In distribution packages we shouldMike Buland
probably include a copy of the final version just for good measure.
2011-06-20Minor tweak, NULL is defined in extratypes.h as per C++ regs if it's notMike Buland
already defined.
2011-06-14With any luck, that'll fix the cross-compiling trace code. Now we just need toMike Buland
hope that the formatter code is not going to have the same problem.
2011-06-08Hopefully nobody was using the buffer size parameter, if so, it's moved.Mike Buland
Buffer now takes an extra parameter: what to buffer, read, write, or both. This winds up being important when dealing with blocking sockets and the like.
2011-05-24Added some empty files that will become my new random number generator system,Mike Buland
also fixed a bug in TcpSocket that caused the system to think the socket was closed when it just connected too quickly.
2011-05-11Tweaked the interface very slightly for insertSorted, very handy change.Mike Buland
2011-05-06Added find operators to list.Mike Buland
2011-04-27The myriadfs program is now in it's own seperate area and not built by default.Mike Buland
2011-04-13Regular expression engine is started, it's...tricky, but I think I can get it.Mike Buland
2011-04-08Rearranged the API a bit.Mike Buland
2011-04-07Pretty sure all utf encoders and decoders are complete and tested.Mike Buland
2011-04-07I fixed a stupid typo in string, I don't know how it ever compiled. It alsoMike Buland
builds on win32 again...
2011-04-06Also realized I had a stupid global macro in place when a template functionMike Buland
would work so much better. It's converted and things are much nicer now.
2011-04-06Tweaked the hash table resizer, it now is more careful about increasing theMike Buland
size of the table when it can reclaim empty space from deletes, and it allows the table to shrink if little enough space is being used.
2011-04-06Libbu++ generates it's own system specific config file as well as a fancyMike Buland
version header file all programs can now use to determine which version, api version, and svn revision of libbu++ they're linking against. It doesn't quite work for windows yet, but it will, eventually.
2011-04-05Fixed the cache system missing a sync on header data for inserts and deletes.Mike Buland
2011-04-04Minor tweak to the cache and cachestoremyriad, fixing a strange corner caseMike Buland
in the cache system.
2011-04-04david - tried to make hash crashDavid
2011-04-04UtfString is going really well. It can now parse Utf8, Utf16 (le,be), andMike Buland
Utf32 (le,be). The internal storage seems to be working fine, although we do have a problem with random access, but at least we can tell which half of a surrogate pair we're on, so we can always rapidly determine the entire code point from any utf16 index that we're on. The only optomization that I'm not doing yet is reading in entire 16bit or 32bit words at a time and converting them from their byte order to native. There are a few potential issues with that, so we'll see. I added a couple of testing datafiles and a test program, I'll delete them all just as soon as it's verified to write correctly.
2011-04-04I made some awesome progress on the UtfString system, it stores in native utf16Mike Buland
encoding to make things easier (little endian in our case). It can currently read utf8 and utf16be, but not BOM. It will give you full unicode code points instead of the raw utf16 values, which is pretty slick.
2011-04-04Really just made some decisions about the overall functionality of the UtfStringMike Buland
and now I'm ready to put some more of the basics into action.
2011-03-30Ok, string stuff is working much, much better, a load of new unit tests haveMike Buland
been added, and I deleted a whole slew of stupid old tests that I don't need.
2011-03-29String's replace function now doesn't get false positives on partial matches atMike Buland
the end of strings. Build should work much better now.
2011-03-27I had a useless include and a silly macro defined in the header file of string.Mike Buland
2011-03-24This may really screw things up, but the trimBack function made no sense, soMike Buland
I changed it. We'll have to add back in the other one if anyone needs it. This will most likely require things to be rebuilt, but they'll complain instead of mislinking this time :-P
2011-03-24The Bu::OptParser no longer tokenizes parameters' values.Mike Buland
2011-03-22Bu::Conduit now works exactly as it was advertised some time ago, it usesMike Buland
Bu::QueueBuf and creates a really slick blocking inter-thread I/O system.