On 01/17/2018 05:27 PM, enh wrote: > On Sun, Jan 14, 2018 at 10:29 AM, Rob Landley <[email protected]> wrote: >> But since out_name is only really ever used in xcreate() (which doesn't >> know that "-" means stdout), and in_name doesn't get modified, this is >> all unnecessary. > > it's not unnecessary. this is the cause of all the test failures you > introduced, including the tar test failures.
There's more than one problem. I removed the FLAG_c setting from gzip_main() because I'd converted do_gunzip(), but I didn't convert do_gzip() and that still needs it... (Sigh. Half-finished...) > can we revert this so i can at least take the other recent changes? > (i'm weeks behind at the moment.) Sorry for the delay answering, my burst of free time ran out and I had to pack up and and drive cross-country for work. I'll try to get this fixed (and passing all the tests) by monday. (I'm also sorry I treated gzip.c as a command in pending. Maybe I should move all the pending commands android's using to the android directory so I don't make that mistake again...) Rob P.S. The gzip tangent I wandered off on before running out of time was generic plumbing to handle file.ext->file and file->file.ext, preferably integrated with the "patch" and "sed -i" file replacement stuff which means rethinking those. I mention it because I deleted it to focus on getting this stuff working again, so hopefully this not will remind me where I left off... P.P.S. Well, the other tangent was writing up proper ps documentation, with more cleanup passes on ps to try to simplify it and make it more understandable. I got maybe the first 1/5 of that checked in even, woo! It was nice to have some free time over the holidays to work on something that wasn't an immediate crisis again... _______________________________________________ Toybox mailing list [email protected] http://lists.landley.net/listinfo.cgi/toybox-landley.net
