On Mon, Apr 3, 2017 at 2:52 PM, Rob Landley <r...@landley.net> wrote:
> // GNU file shows the source OS and timestamp too but busybox and toybox > // leave those unknown/blank, so reporting them isn't obviously useful. > > Busybox hasn't got "file". I agree reporting them isn't obviously > useful, but I think I'm yanking the comment (or at least moving it to > the checkin comment). > i meant "busybox and toybox [when generating gzip data] leave those unknown/blank". > And while we're at it, in zip detection... > > int ver = (int)(char)(toybuf[4]); > > A double typecast is never a good sign, but here we compile with > -funsigned-char (for consistency across architectures) which toybuf[4] > already _is_ an instance of so... what is this doing? Did they mean to > typecast it to (signed char) and then... llvm complained maybe? Can the > zip version be negative? > > I'm confused! me too. the rest of the code uses peek_le/peek_be for that kind of thing. 830f2561 toys/pending/file.c (Isaac Dunham 2016-02-29 11:29:21 -0600 247) int ver = (int)(char)(toybuf[4]); > > Rob > -- Elliott Hughes - http://who/enh - http://jessies.org/~enh/ Android native code/tools questions? Mail me/drop by/add me as a reviewer.
_______________________________________________ Toybox mailing list Toybox@lists.landley.net http://lists.landley.net/listinfo.cgi/toybox-landley.net