trinque: alright, turns out it's not so bad to have the chmod +x happen in my build scripts.
trinque: operator will have a top-level dir of scripts to chmod +x after press (and it seems appropriate for this to be done with human hands and decisions), and the rest will flow from that.
trinque: or hell, can run them with `sh script` for all I care
shinohai: now thats a good idea trinque (to have top-lvl dir of scripts one expects to set +x so as not to have to manually locate)
asciilifeform: trinque: imho nuffin wrong w/ e.g. set_permissions.sh that restores the magick bits after press.
asciilifeform: (or with 'sh script' for that matter )
jfw: I agree (?) that vdiff isn't broken here; it's established behavior of diffs to treat empty files & directories as nonexistent, indeed this is relied on by patch -E / vpatch to know what to delete. The empty files and mode bits are 'impedance mismatches' between v and git, where the tool people use shapes their ideas of reasonable expectations
jfw: ah, I may be mistaken about 'established behavior', traditional diffs may use the timestamp field to distinguish, but vdiff can't as it omits the field.
asciilifeform: jfw: my original vdiff.sh kept timestamps. i cut'em out shortly after init release, operating on traditional pos. (endorsed by mp, and i agreed) that polit. calendar time is bogus.
asciilifeform doesn't miss the timestamps; and has quite a few boxes w/ questionable clocks where they'd end up ~random #s
trinque: jfw: I am disinclined to take precedent for anything here, but am satisfied that "supporting empty files" is pointless
trinque: and also file modes