dstailey@gnu.ai.mit.edu (Doug Stailey) (12/21/90)
Is anyone else trying to port bash to MiNT? I have a version ported now that is somewhat flawed, but still works. The problem lies in bash's interpretation of control-Z, which stops jobs, but fails to register them in the job queue. Processes or pipes started in the background, using the ampersand, show up when you type "jobs," but processes stopped with control-Z do not. The fg & bg builtins do not work with processes stopped with control-Z, but the fg & bg utilities provided with MiNT will work, as will the MiNT version of kill. I am using patchlevel four of the MiNT libs. Bash runs in raw mode, so the interpretation of control-Z is up to it. I am using the source code for version 1.05, using some of the modifications my brother Ken added to his port of bash to Minix. (Herein may lie one of the main problems with control-Z, in something that was done to glob.c. I am looking into this as a possibility, but bash's source is astronomically large, and the true cause may take a week or two to find.) Outside of this problem, bash is working out really well. I just had three invokations working as the default shells under MGR. I have modified bash to know that it is the login shell when it is invoked as init.prg, and it tries to read ~/init.rc. .bash_history is now history.sh, and the path separator is a comma. Because of the control-Z problem, I consider this port far too flawed to even suggest releasing. However, if you are as desparate for a MiNT version of bash as I am, I will email copies on request. You may want to hold off though, since I am working hard on a solution. -- disclaimer: This message is sold by weight, not volume. Contents may have settled during shipment.