Skip to content

HTTPS clone URL

Subversion checkout URL

You can clone with HTTPS or Subversion.

Download ZIP
Newer
Older
100644 120 lines (85 sloc) 4.463 kb
4472db9 Refactor: make bgio use globals again. Sine the user only interacts wit...
Scott Bronson authored
1 - Get rid of master_pipe. Spread functionality into tasks, and the app
2 should just keep a global of the current task.
3
641d8e3 Make idle modular so it can be used by any command task (rz, rsync, etc)
Scott Bronson authored
4 - We currently freeze when the transfer is cancelled. That's bad.
5
6 - rzh -q: tell current rzh to quit
7 (doesn't matter how deep into other machines you are; this bangs
8 you right back up to your most recent rzh).
9 - rzh -u: return to the next machine upward. ^D or rzh -q to move downward.
10 - rzh -i: tell where downloads are currently going
11 (not envar hack; literally hit the terminal to find out)
12
e9de567 Update docs, some tests, add PRODUCTION=1 to Makefile.
Scott Bronson authored
13 - Add listening for keypresses to idle.c.
641d8e3 Make idle modular so it can be used by any command task (rz, rsync, etc)
Scott Bronson authored
14 - Ensure we make the idle string the right width -- always.
098984c In-progress checkin; remove --verbose, lots of cleanups to test director...
Scott Bronson authored
15
16 - Add a real message for when a file is skipped.
17 Should recommend using "sz -y" or "sz -N" to ensure file is sent.
18
19 - Could parse the zmodem stream to discover the filename and size
20 so we can provide a better in-progress display.
21 Actually, play around with rz -v -v -v first.
22
e9de567 Update docs, some tests, add PRODUCTION=1 to Makefile.
Scott Bronson authored
23 - Release 1.0
24
25 - Need to worry about overflowing byte counters. Convert them to long longs?
26 Otherwise the numbers will go wrong after 4 GB transferred.
098984c In-progress checkin; remove --verbose, lots of cleanups to test director...
Scott Bronson authored
27
641d8e3 Make idle modular so it can be used by any command task (rz, rsync, etc)
Scott Bronson authored
28 - Is there any way to get the shell to flush its history? Right now,
29 starting rzh screws up history (hit up arrow, down, run rzh, hit up
30 arrow -- notice how the history items are different).
098984c In-progress checkin; remove --verbose, lots of cleanups to test director...
Scott Bronson authored
31
4bc39ec note refactor branch
Scott Bronson authored
32 20 Apr 2006:
33 * Branch 0.8 just before the big scary refactoring.
9577bf9 Added preflights to check for rz executable and writable dstdir.
Scott Bronson authored
34
098984c In-progress checkin; remove --verbose, lots of cleanups to test director...
Scott Bronson authored
35 20 Mar 2006:
e9de567 Update docs, some tests, add PRODUCTION=1 to Makefile.
Scott Bronson authored
36 * Finally got the last of the known transfer bugs. It now works pretty well!
098984c In-progress checkin; remove --verbose, lots of cleanups to test director...
Scott Bronson authored
37 * Bump version to 0.8.
38fa969 Add RZHDIR environment variable.
Scott Bronson authored
38
39 17 Jun 2005:
098984c In-progress checkin; remove --verbose, lots of cleanups to test director...
Scott Bronson authored
40 * It works! It finally sometimes works.
49aaf0d Big reorg is mostly done. Echo works, rz still needs to be debugged.
Scott Bronson authored
41
41efe92 tiny update to changes file
Scott Bronson authored
42 13 Jun 2005:
d2dc9e6 We now terminate the child when we see it send a ZFIN. This means
Scott Bronson authored
43 * Try forking a new rz process to handle the zmodem transfer.
098984c In-progress checkin; remove --verbose, lots of cleanups to test director...
Scott Bronson authored
44 pdpzm works but it has some boundary case bugs and is excruciatingly slow.
41efe92 tiny update to changes file
Scott Bronson authored
45
8500151 Forgot to disable the bad tests.
Scott Bronson authored
46 13 Nov 2004: tagged version 0.2 (r27)
2e7caf7 Version 0.2, second version sent to Paul Edwards.
Scott Bronson authored
47 * Made zmfr.c modular, sent in patches.
48 * Added send functionality but it appears very broken.
49
098984c In-progress checkin; remove --verbose, lots of cleanups to test director...
Scott Bronson authored
50 05 Nov 2004: tagged version 0.1 (r21)
d2dc9e6 We now terminate the child when we see it send a ZFIN. This means
Scott Bronson authored
51 * Switched to the pdpzm library because the lrzsz code is amazingly brittle.
41efe92 tiny update to changes file
Scott Bronson authored
52
d891589 First attempt at converting rzh off off pdpzm. There's a big bug in ech...
Scott Bronson authored
53 15 Jan 2004:
d2dc9e6 We now terminate the child when we see it send a ZFIN. This means
Scott Bronson authored
54 * First version. Tried to modify lrzsz's rz command to add this feature.
16efd6d rzh 0.11: the first version I sent to Paul Edwards
Scott Bronson authored
55
56
e9de567 Update docs, some tests, add PRODUCTION=1 to Makefile.
Scott Bronson authored
57
58 REFACTOR:
59
60 There are problems when cancelling the transfer using ^C. Some of the
61 data hits the screen.
62
63 Why does this command stall: ./fireup Makefile < /dev/null
64 Probably because of shoddy i/o. Need to refactor first.
65
66 todo: write some functional tests.
67 todo: set all buffers to 1 byte and see if we garble anything
68 todo: set all buffers to 7 bytes (1 larger than start string)
69 and see if we garble anything
70
71 - It's ridiculous to have the fds and pids in the task spec.
72 Get rid of them, then make all the task specs static. Those are
73 the classes. Then, pass the fds and pids to task_install to
74 create the instances.
75 This will get rid of the duplicate closes.
76
77 - Should turn master into just another task. Only difference between
78 master and regular tasks is that master got added first.
79
80 - Move logio into the log library. Need a way for non-fifo parts of
81 the code to log buffer contents. Clean up logio in zrq.c.
82 Call it log_buffer or log_data or something.
83
84 Maybe never: Make --rz and --shell work.
85 todo: add specifying command for to use (-c).
86 (this should help testing because I could do "rzh -c sz ..."
87
88 TODO: rzh --up, rzh -u, rzhu, or rzu: move up
89 TODO: rzh from rsync.
90 TODO: rzh from tar.
91
92
93
098984c In-progress checkin; remove --verbose, lots of cleanups to test director...
Scott Bronson authored
94 LONG TERM:
95
96 - Fifo is overcomplex. Should have used shiftbufs so I don't packetize
97 the stream more than I should. Switch fifo.c to shiftbuf.c?
98 No. Instead, finish the netknife infrastructure, then switch it over
99 to that.
100
101
102 NO NEED:
49aaf0d Big reorg is mostly done. Echo works, rz still needs to be debugged.
Scott Bronson authored
103
104 todo: make expandable fifos so masses of memory will guard against deadlock.
098984c In-progress checkin; remove --verbose, lots of cleanups to test director...
Scott Bronson authored
105 Actually, don't do this. There doesn't seem to be a need.
106 Especially because zmodem doesn't ever read, it only writes!
d18f1b8 Implement the -q flag (finally), make the idle proc smarter about when
Scott Bronson authored
107
49aaf0d Big reorg is mostly done. Echo works, rz still needs to be debugged.
Scott Bronson authored
108 todo: there is a slight chance of deadlock when we're installing a new task.
109 The problem is draining the fifo for the new task. Instead, we should
110 mark the fifo, and give it the new task to install. Then when the fifo
111 has drained down to that point, we'll kick it over to the new task
112 automatically. Right now, fifo_drain is really, really, evil.
113 NOTE: this should solve the nested task create / sigchild issue.
098984c In-progress checkin; remove --verbose, lots of cleanups to test director...
Scott Bronson authored
114 (this issue is where we receive a sigchld when the buffer contains
115 data intended for the rz task and data intended for the outer shell.
116 Who handles the sigchld?)
117 As it turns out, the deadlock never even comes close to happening and
118 it's easy enough just to handle the sigchld late. Sure, it would be
119 more correct to mark the buffer, but this is good enough for now.
Something went wrong with that request. Please try again.