| 1 | Revision history for Perl extension AnyEvent. |
1 | Revision history for Perl extension AnyEvent. |
| 2 | |
2 | |
| 3 | TODO: strict checks for file, and valid fd |
3 | - actually make the signal pipe work under win32. |
| 4 | 4.341 |
4 | - localise $! in signal handler to avoid changing $!, although |
|
|
5 | perl itself does not seem to save/restore errno either. |
|
|
6 | - set the cloexec flag on the signal pipe (normally set by perl too). |
|
|
7 | |
|
|
8 | 4.35 Fri Mar 27 11:48:20 CET 2009 |
| 5 | - event models relying on AnyEvent's signal watcher emulation did |
9 | - event models relying on AnyEvent's signal watcher emulation did |
| 6 | invoke the callback asynchronously, contradicting documentation |
10 | invoke the callback asynchronously, contradicting documentation |
| 7 | and causing signals to get lost (this includes AnyEvent's own |
11 | and causing signals to get lost (this includes AnyEvent's own |
| 8 | event loop). AnyEvent now uses the standard pipe trick to make |
12 | event loop). AnyEvent now uses the standard pipe trick to make |
| 9 | callback execution synchronous to the event loop. |
13 | callback execution synchronous to the event loop. |
| 10 | - AnyEvent::Handle didn't free tls context data on DESTROY |
14 | - AnyEvent::Handle didn't free TLS context data on DESTROY |
| 11 | (patch by Pavel Shaydo). |
15 | (patch by Pavel Shaydo). |
| 12 | - work around the ever-incompatibly-changing API of MakeMaker. |
16 | - work around the ever-incompatibly-changing API of MakeMaker. |
| 13 | - document that changing global variables without restoring them |
17 | - document that changing global variables without restoring them |
| 14 | is a bad idea in Perl (noted by Adam Rosenstein). |
18 | is a bad idea in Perl (noted by Adam Rosenstein). |
|
|
19 | - AnyEvent::Strict now barfs if ->io is passed a file. |
| 15 | |
20 | |
| 16 | 4.34 Thu Feb 12 18:32:45 CET 2009 |
21 | 4.34 Thu Feb 12 18:32:45 CET 2009 |
| 17 | - separately buffer TLS read data, as otherwise the read queue could |
22 | - separately buffer TLS read data, as otherwise the read queue could |
| 18 | deadlock as receiving data is not expected while draining the |
23 | deadlock as receiving data is not expected while draining the |
| 19 | read queue (which cna only happen with TLS). |
24 | read queue (which cna only happen with TLS). |
