Doug Coleman [Fri, 4 Mar 2016 00:53:24 +0000 (16:53 -0800)]
command-line.startup: document more command line flags. we already have -nosignals, but it was hidden in the source code. Fix spacing for default image. Closes #1514.
All versions of SSL are deprecated and have major security issues. Just
kill the whole thing. If you want to deliberately expose yourself to
POODLE, you can do it on your own time.
Benjamin Pollack [Tue, 23 Feb 2016 13:09:07 +0000 (08:09 -0500)]
openssl: only set RSA keys if required to do so
Modern OpenSSL and LibreSSL both do not require SSL_CTX_set_tmp_rsa to
be called unless SSL_CTX_need_tmp_rsa returns true, and LibreSSL and
OpenSSL compiled with deprecation warnings both will fail if this
happens. This commit resolves that.
With this change, it becomes possible to use LibreSSL in place of
OpenSSL with Factor.
Benjamin Pollack [Mon, 22 Feb 2016 18:37:54 +0000 (13:37 -0500)]
Update OpenSSL calls to work with LibreSSL/newer OpenSSL
RSA_generate_key has been deprecated, and is not available in either
LibreSSL or newer OpenSSL that are not compiled with deprecated call
support. This commit switches to the still-supported
RSA_generate_key_ex, which is supported by both libraries.
Note that this is still insufficient for Factor to work under LibreSSL,
though we now get a call further than before.
Benjamin Pollack [Thu, 18 Feb 2016 19:29:54 +0000 (14:29 -0500)]
Add support for LibreSSL's libtls library
These are just the low-level FFI bindings; high-level bindings (and
maybe an alternative implementation of io.sockets.secure) will come in
a later commit.
Benjamin Pollack [Wed, 17 Feb 2016 17:23:09 +0000 (12:23 -0500)]
Fix incorrect alien.syntax documentation for FUNCTION:
The library docs indicate that you need a semicolon at the end of
a `FUNCTION:` definition, but that's actually a syntax error. Likewise,
while the parentheses and commas may at one point have been optional,
they emphatically aren't anymore. This patch brings the documentation in
line with the code.
Jon Harper [Sat, 9 Jan 2016 18:06:26 +0000 (19:06 +0100)]
x11 xlib, use Xutf8LookupString instead of XwcLookupString
XwcLookupString uses a locale dependant encoding, but was always
decoded using utf16. On my system, it was not utf16 that was used.
Xutf8LookupString always uses utf8 so it should work everywhere
Björn Lindqvist [Thu, 17 Dec 2015 17:57:52 +0000 (18:57 +0100)]
benchmark: wrap the error in <test-failure>
Piggybacks on the test-failure error. word so that we get a traceback if
an error occurs during benchmarking. To make it easier to debug the "No
suitable arithmetic method" random error #1484
Björn Lindqvist [Wed, 9 Dec 2015 20:31:13 +0000 (21:31 +0100)]
VM: put the singletons t, -1, 0 and 1 in the special objects table
Having them there instead of as fields in the vm struct removes a bunch
of special handling. This commit just adds them and doesn't remove the
old ones to avoid potential chicken-and-egg bootstrap problems.
Björn Lindqvist [Thu, 3 Dec 2015 11:00:40 +0000 (12:00 +0100)]
tools.test: print the callstack in test failures
So when mason reports a test failure it should include the whole
callstack. That should make it easier to catch the "No suitable
arithmetic method" random error that is hard to debug without a
callstack.
Björn Lindqvist [Fri, 4 Dec 2015 12:29:29 +0000 (13:29 +0100)]
bootstrap.*: fix for #1512, load docs after everything else
By making it so the help component is loaded almost last, all vocabs are loaded without docs and only then are they loaded. It should make it so you don't get cyclic dependency problems in doc files during bootstrapping.
Björn Lindqvist [Thu, 3 Dec 2015 23:57:19 +0000 (00:57 +0100)]
ui.backend.gtk: set the active? to 100 in on-map, should fix #1506
what now happens is that configure events are skipped until the first
on-map signal is fired. seem to work fine in a variety of wm:s. if this
works well we can return active? to be plain boolean.