On Sun, Apr 02, 2006 at 11:08:27AM +0100, Julian Gilbey wrote: > > > BTW, I have an additional way of avoiding the "killing the BTS at > > > 02:00 local time": just change the heading to say: > > > > > > # Best-practice information for Debian developers > > > # intended to be run periodically as a cronjob, for example weekly as: > > > # 0 hour * * day /usr/bin/dev-best-practice > > > # (replacing hour and day by your preferred times!) > > I prefer to give something copy-and-paste-able, and which doesn't > > assume that everyone wont use day=0 hour=0. > > There are two options here: > (1) Trust that developers are competent enough to make the change > themselves, or > (2) Not trust them. > > And everyone using day=0 hour=0 will be identical to everyone using > day=0 hour=2! Er, I meant 2..
But I'm not sufficiently creative to come up with times more random than 01:23 :/ > > # 0 2 * * 0 /usr/share/doc/devscripts/examples/dev-best-practice --delay > > --mail > Need to change this line... Do you mean about moving to /usr/bin/? > > while test $# -gt 0; do > Use getopt :) Ugh. afaics all the getopt foo requires looping anyway, so I just did it myself. > > if [ -n "$XMAIL" ]; then > > exec 3>&1; > > exec 1> >(mail -s "Debian fixme list for $LOGNAME on `date +%x`" > > $LOGNAME) > Nice one :) There's also exec >& >(tee -a "$0.log"); to make a self-logging script, or pdebuild >& (tee sextractor.log) Or, a 1-liner to implement annotate-output: ls >& >(while read l; do echo "`date +%X`: $l"; done) > > echo "Checking for your bugs tagged moreinfo:"; > > lynx -dump "$bts/from:$DEBEMAIL&include=moreinfo" | > > I'd prefer not to rely on lynx/links if possible and make do with > wget/sed/perl. Then you need something like: wget -qO- 'http://bugs.debian.org/from:[EMAIL PROTECTED]&include=moreinfo' |grep '^<li><a href="bugreport\.cgi?bug=[^"]*">' |cut -d'>' -f3 |cut -d'<' -f1 lynx -dump is not wget -qO- .. this leaves crud like " in the bug titles. It both makes fragile assumptions about the BTS format, and mangles bug titles. > > sed -ne '/^[[:space:]]*\* \[[[:digit:]]\+\]\(#[[:digit:]]\)/s//\1/p'; > > # grep -E '^[[:space:]]*\* \[[[:digit:]]+\]#[[:digit:]]+' | > > # sed -e 's/^[^#]*//'; > > Why the commented out lines? It's an alternate implementation, which I haven't yet discarded. > > if [ -n "$XMAIL" ]; then > > exec >&-; > > exec 1>&3; > > fi; > > Not sure what this bit does; wouldn't exec 1>&3- be adequate? You're probably right. > Finally, I think it would be nice to have a devscripts option > DEVCHECKER_EXCLUDE which would be a colon-separated list of tests to > skip. Then each test would read: > > case ":$EXCLUDE:" in > :thistest:) # skip test > ;; > *) # do test > [...] > ;; > esac I don't understand; are $EXCLUDE and $DEVCHECKER_EXCLUDE distinct? Do you mean to split with IFS=':', and loop around it for each test? Why not separate each check as a function, and list that function in $tests, and loop: for test in $tests; do grep -q ":$test:" <<<":$DEVCHECKER_EXCLUDE:" && continue; $test; done; Justin -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
