| Change | User | Description | Committed |
|---|---|---|---|
| 19936 | mbi | Making an internal dictionary concurrent because reads aren't lock protected and if you ru...n a lot of commands eventually it'll resize during a read and everyone will be sad. This fixes an issue where a managed callback invoked by native code could throw an exception, which gets discarded and a generic error message produced because native can't do anything useful with it. « |
9 years ago |
| 19935 | mbi | change LastResults thread protection to use a reader/writer lock instead of interlocked in...teger guard, which is much faster and safer than my previous change. « | 9 years ago |
| 19934 | mbi | adding the ability to specify the encoding type of command responses sent from the server.... This is a server-wide setting that is only used for text results of p4 commands. I also removed some code that would prevent the use of non-ascii encodings, even if you explicitly called the utf8 or utf16 function. Also, cleaning up the UTF8 marshal helper a bit. « |
9 years ago |
| 16708 | mbi | Trying to improve last results cache behavior. This isn't an ideal solution but it fixes... a couple race conditions related to many threads periodically running commands having their results get cleaned up by each other while they are accessing them. « |
9 years ago |
| 16707 | mbi | Adding a call to explicitly evict last results so that commands with large result sets don...'t hang around in memory forever. This is a concern for applications which have only a small number of threads which ever access perforce, as the results get orphaned at the termination of the thread and never cleaned up unless >32 distinct threads run commands. « |
9 years ago |
| 16486 | mbi | Resolving an issue that can occur when two threads simultaneously attempt to clear the las...t results cache of stale results. « | 9 years ago |
| 16481 | mbi |
Branching //guest/perforce_software/p4api.net/... to //guest/mbi/p4api.net/... |
9 years ago |