Many hyperlinks are disabled.
Use anonymous login
to enable hyperlinks.
35 check-ins using file config.c version 8c0bacf82f
|
2003-04-25
| ||
| 12:44 | The new ssh_do_close() function itself causes a segfault if the connection goes foom before ssh->channels is set up. Oops. Fixed. check-in: 899cb52b70 user: simon tags: trunk | |
| 11:42 | Fix another segfault on abrupt X connection shutdown. This should have happened in rev 1.5 [r996] but didn't! Now we never call sk_get_private_ptr() on a socket unless we've ensured it's non-NULL. check-in: 738c707e42 user: simon tags: trunk | |
| 10:44 | Don't block SIGPIPE until _after_ pterm forks, since otherwise the blockage will propagate to the child process. check-in: 9546ad05f2 user: simon tags: trunk | |
| 06:48 | Robustness in the face of sudden connection closures: we now make a credible effort to shut down open forwardings cleanly when the owning SSH connection terminates abruptly (for whatever reason). check-in: 26f97a0d19 user: simon tags: trunk | |
|
2003-04-23
| ||
| 10:25 | Force gcc to use x86's built-in 64/32->32,32 division instruction rather than swanning off to a helper function. This seems to make woefully little actual performance difference, which annoys me, but it's a just-about-detectable improvement so I'll leave it in. check-in: 0aa2c11554 user: simon tags: trunk | |
| 09:48 | Added framework to sshbn.c to make it possible to vary the underlying integer type forming the Bignum. Using this, arranged that gcc/x86 uses 32-bit chunks rather than the guaranteed ANSI- portable 16-bit chunks. This has gained another 30% on key exchanges by my measurements, but I'm not yet convinced that it's all perfectly robust - it seems to work fine for SSH1 and SSH2/RSA but I haven't ensured that every bignum routine is actually being tested, so it may yet show up problems in DSA or key generation. check-in: 3769f9433e user: simon tags: trunk | |
| 08:48 | Unix makefile now uses -O2, which massively cuts down key exchange time. This gives rise to a whole bunch of spare warnings, one or two of which might have been actual bugs; now all resolved. check-in: 6c10570c0b user: simon tags: trunk | |
|
2003-04-18
| ||
| 08:09 | Missing bits for colour selection support. check-in: 95a72b4829 user: ben tags: trunk | |
| 08:09 | Missing bits for System 7 listbox support. check-in: acc262a072 user: ben tags: trunk | |
| 04:14 | Removing items from a list box using gtk_container_remove is nasty, because when the selected item is removed the selection moves on to another item. Thus, calling dlg_listbox_clear causes repeated selchanges in the list, which in turn cause repeated valchanges if the list is attached to a combo box. This has been completely scuppering the Translation panel. check-in: a530f9d4d5 user: simon tags: trunk | |
| 04:06 | Oops; we _do_ need to set dport even when doing dynamic forwarding, because otherwise the `if (sport && dport)' statement runs the risk of not working. check-in: f3060f0139 user: simon tags: trunk | |
| 04:00 | Missed out a couple of uxsel_tell() which were hanging forwarded X connections in at least some situations. Oops. check-in: 384eaf1a25 user: simon tags: trunk | |
| 04:00 | Fix obvious foulup in X forwarding: my recent `how did this ever work' change to port forwarding completely broke it, by failing to set peeraddr{,len} before reading them. Oops. check-in: 58503ab138 user: simon tags: trunk | |
|
2003-04-16
| ||
| 18:58 | Replace use of FIONBIO with POSIX-approved O_NONBLOCK (set using fcntl()). This should save us having to worry about different OSs' defining it in different headers. check-in: bc49ee811b user: ben tags: trunk | |
| 18:33 | Make IPv6 support for Unix work, and make it a lot simpler in the process. At the moment, we have to assume that getaddrinfo() will only return AF_INET and AF_INET6 addresses, since we patch in the port number into the sockaddr later. Fixing this is probably best done by redesigning the PuTTY network abstraction a little. check-in: 6206b1162a user: ben tags: trunk | |
|
2003-04-14
| ||
| 18:47 | Remove all those horrid fprintfs. check-in: 91ec589e47 user: ben tags: trunk | |
| 18:28 | Set the space required by a listbox sensibly as well. check-in: 3c809505f1 user: ben tags: trunk | |
| 18:25 | Minor fixes: - Erase the background of System 7 editboxes before drawing them. - Set the height of listboxes correctly. - Set the panel-selection listbox to 20 items high. check-in: 82e19cbda1 user: ben tags: trunk | |
| 17:55 | Don't call GetColor() if we don't have Color Quickdraw -- no good will come of it. check-in: f093d4afaa user: ben tags: trunk | |
| 17:42 | Support for colour selection in the Mac OS. check-in: 1bb87f2427 user: ben tags: trunk | |
| 16:34 | Keyboard events in Mac OS 8 list boxes should generate EVENT_SELCHANGE, not EVENT_VALCHANGE. check-in: 0d697e5d61 user: ben tags: trunk | |
| 16:15 | Enlarge the settings window so as to fill a 640x480 display. This makes enough space to actually fit all of the panels. I'll have to do a version with smaller fonts for 9" screens. check-in: c660bc8f15 user: ben tags: trunk | |
|
2003-04-13
| ||
| 09:37 | Add a list box to the config dialogue to switch between panels. Now I don't have to recompile to change panels! check-in: 22c49c931e user: ben tags: trunk | |
| 08:52 | Support for list boxes under System 7. There are some horrible kludges in this code to do with the handling of the scroll bar, but it doesn actually work better than the Mac OS 8 version. check-in: 6af1a1874e user: ben tags: trunk | |
|
2003-04-12
| ||
| 16:23 | socks5_negotiate(): rather than using an uninitialised value when passed an address of an invalid type, assert() that it will never happen. Do something similar for SOCKS4 as well. check-in: 125ea26ede user: ben tags: trunk | |
| 16:15 | Don't use an uninitialised value when we get an unsupported ATYP in a SOCKS5 request. Spotted by GCC. check-in: 0bc01ae5f3 user: ben tags: trunk | |
| 16:07 | Remove a couple of spurious semicolons. check-in: 2e726f9adc user: ben tags: trunk | |
| 16:06 | Update for Simon's from_backend changes. check-in: 099db98855 user: ben tags: trunk | |
| 12:37 | Support for New Session, Duplicate Session and the Saved Sessions submenu in Unix PuTTY, and Duplicate Session also in pterm. You do _not_ want to know how this is done. Be warned. check-in: 2a16a8e814 user: simon tags: trunk | |
| 06:03 | Further fiddlings with the size reconfiguration stuff; now everything seems to work right even when the only thing changed was the window border. check-in: b3bd56e4fb user: simon tags: trunk | |
| 05:44 | Mid-session reconfiguration of scrollback was failing (in Unix) except when width or height had _also_ been changed. Fixed. check-in: 8d3c66813d user: simon tags: trunk | |
| 04:27 | Implement the ldisc `special' mechanism in Unix, without which local line editing didn't work at all. check-in: 82c66f5741 user: simon tags: trunk | |
| 04:19 | Fix from yesterday's frontend-handle upheaval: ldisc calls from_backend(), and must now pass its frontend handle rather than its terminal handle. check-in: ea2283da3c user: simon tags: trunk | |
| 04:05 | Mid-session BCE reconfiguration wasn't taking effect immediately because I forgot to set term->erase_char in response to the change in term->use_bce. Oops. check-in: 32701df063 user: simon tags: trunk | |
| 03:59 | Disable window title reporting by default, since it's a security hazard. I considered removing it completely, but I can't rule out the possibility of an OS that actually takes security of its terminal devices seriously, and which might be able to make sensible and safe use of this feature. check-in: a301bcba32 user: simon tags: trunk | |