Okay, I changed the code so that no other thread tries to access the GUI (a lot of global variable stuff, but this doesn't matter at the moment). When starting more than one thread, the GUI shouldn't freeze anymore (I tested it several times and it runs well, even on Windows). GThread was totally removed. Would be nice if you and others could test it.
Furthermore, I fixed a bug concerning the charset: when no charset.txt was found in the folder, it will use a hardcoded one. Maybe I will add a possibility to specify a custom charset file in the future, we will see.
At last, I added a 'Do you really want to quit?' dialog which pops up when the user quits while rcracki_mt is still running. If this is the case, it will keep the session files on disk.
In addition, there are several smaller things I changed/fixed: you can now browse where to save the cracked hashes and the session files, the progress bar has a percent indicator, verbose output etc.
Of course there are some minor bugs, and I don't have a clue why you can't mark any text, quel, but I think a 'Clear Messages' and a 'Copy result to clipboard' button would be useful. I also tried to compile the executeable with static libraries, but static libraries and GTK+ is a pain (lots of undefined references). However, here's the new version, just type 'make' and run rcracki_mt_gui.
First, again thank you! Second, at the rate you are going minimally you should sign up for gitorious and clone the project so you can keep commit history of changes and eventually issue a pull request (http://gitorious.org/freerainbowtables- ... nbowtables
). It is currently synced with rcracki.sourceforge.net trunk (64-bit fixes for overflows mostly compared to 0.6.5.2.) If you prefer svn then I'll have to see if I can create a branch where you have commit access.
1) an option to just input one hash instead of loading a file would be nice
2) thanks for the more flexible result saving option (and session options)
3) pausing during pre-calculation seems to work sometimes but not always ie it finishes that pre-calculation but then keeps going instead of pausing like the cli does
4) if we're in a paused state shouldn't we be able to abort?
5) sometimes abort, even with 1 thread, doesn't seem to have impact and after 20 seconds a second abort does (especially true if I pause, resume, then keep clicking abort)
6) test a variety of keep precalc, abort, pause, resume, start options and they all seemed to work correctly minus #5
7) memory limit set to anything including 50 seems to still get me 541mb+ memory (ie the size of a table file plus some index)
8) the hardcoded charset is one of those double edged swords - until we have complex hybrids keeping it as a file gives a lot of flexibility. Also, with RTI2 the charset will be part of the file header so there won't be a need for this hardcode.
9) I can mark text in single and multi-threaded thread mode (and copy it)
10) copy result to clipboard is probably a good idea
11) clear is also a good idea in case people do multiple runs
12) still need some way to wildcard table inputs
13) I see your point about a static build being obnoxious
14) If you take a look at https://build.opensuse.org/project/pack ... %3Aquelrod
you may be able to get some compilation/build testing for a decent variety of GNU/Linux systems. At least to know what libraries we might be missing.