Mercurial > pidgin.yaz
view autogen.sh @ 3752:b32474e522fa
[gaim-migrate @ 3890]
From: "William T. Mahan" <wtm2@duke.edu>
This patch, against CVS HEAD, fixes three bugs in Oscar File Transfer
support. I can split it up further if desired.
* Send a null checksum when initiating a file transfer, which fixes
"files don't match" warnings produced by some versions of WinAIM; add
a compile-time option to actually compute the checksum, which is
slow but necessary when sending to some Mac clients.
* Don't allow sending files to oneself, because it causes all kinds of
subtle problems and it's not useful.
* Don't crash when there is an error writing to the output file when
receiving.
From: "William T. Mahan" <wtm2@duke.edu>
This patch 2 of 3, which applies on top of the first, adds support for
reverse connections for Oscar File Transfer, the lack of which has
been the biggest complaint so far. Reverse connections are used by
newer AIM clients when there is difficulty verifying the IP of the
sender.
From: "William T. Mahan" <wtm2@duke.edu>
This patch 3 of 3, which applies on top of the first 2, removes the
alarm() and sigaction() calls that were added by my original FT patch
to detect transfer timeouts. Besides apparently not working on
Windows, they involved a lot of ugly code to handle a special case.
My new approach is to add destructors that can called when SNACs are
freed; a timeout is detected when a request SNAC is cleaned up before
the transfer is accepted. Although this touches several files, it is
more generic than the old method. I tried to implement this in an
unintrusive manner, so that there is little preformance penalty for
SNACs that do not use destructors.
My first two patches should work fine without this. If there are any
objections to the third patch, I ask that the first two patches be
applied, in which case I will set up a SourceForge page for this one.
committer: Tailor Script <tailor@pidgin.im>
author | Luke Schierer <lschiere@pidgin.im> |
---|---|
date | Sat, 19 Oct 2002 05:22:30 +0000 |
parents | 1496be1c345c |
children | 27354602734d |
line wrap: on
line source
#!/bin/sh SETUP_GETTEXT=./setup-gettext ($SETUP_GETTEXT --gettext-tool) < /dev/null > /dev/null 2>&1 || { echo; echo "You must have gettext installed to compile Gaim"; echo; exit; } (libtoolize --version) < /dev/null > /dev/null 2>&1 || { echo; echo "You must have libtool installed to compile Gaim"; echo; exit; } (automake --version) < /dev/null > /dev/null 2>&1 || { echo; echo "You must have automake installed to compile Gaim"; echo; exit; } (autoconf --version) < /dev/null > /dev/null 2>&1 || { echo; echo "You must have autoconf installed to compile Gaim"; echo; exit; } # Thanks decklin if test -f configure.ac ; then if autoconf --version | grep '2\.[01]' > /dev/null 2>&1 ; then mv configure.ac configure.2.1x; echo "configure.ac has been moved to configure.2.1x to retain compatibility with autoconf 2.1x" echo "Future versions of Gaim will not support autoconf versions older than 2.50" fi fi echo "Generating configuration files for Gaim, please wait...." echo; # Backup the po/ChangeLog. This should prevent the annoying # gettext ChangeLog modifications. cp -p po/ChangeLog po/ChangeLog.save echo "Running gettextize, please ignore non-fatal messages...." $SETUP_GETTEXT # Restore the po/ChangeLog file. mv po/ChangeLog.save po/ChangeLog echo "Running libtoolize, please ignore non-fatal messages...." echo n | libtoolize --copy --force || exit; # Add other directories to this list if people continue to experience # brokennesses ... Obviously the real answer is for them to fix it # themselves, but for Luke's sake we have this. for dir in "/usr/local/share/aclocal" \ "/opt/gnome-1.4/share/aclocal" do if test -d $dir ; then ACLOCAL_FLAGS="$ACLOCAL_FLAGS -I $dir" fi done aclocal -I m4 $ACLOCAL_FLAGS || exit; autoheader || exit; automake --add-missing --copy; autoconf || exit; automake || exit; ./configure $@