It appears that bugs.debian.org has changed the CSS and HTML tags which is what is causing debchange.pl to get the wrong information when parsing the HTML.
This regex no longer matches:
675 if (m%<a(?: class=\".*?\")?
href=\"bugreport.cgi\?bug=([0-9]*).*?>\#\1: (.*?)</a>%i) {
Inserting a simpler regex (print $_ if (/bugreport.cgi/);) shows that
the actual HTML is:
$ /opt/debian/src/devscripts/devscripts-2.10.12/scripts/debchange.pl --closes
459945
<li><a href="/cgi-bin/bugreport.cgi?bug=459780">#459780: emdebian-tools:
emsource uses "/" as working directory by default</a><li><a
href="/cgi-bin/bugreport.cgi?bug=459796">#459796: [INTL:gl] Galician debconf
template translation for emdebian-tools</a><li><a
href="/cgi-bin/bugreport.cgi?bug=459867">#459867: emdebian-tools: [INTL:eu]
Basque debconf templates translation update</a><li><a
href="/cgi-bin/bugreport.cgi?bug=459945">#459945: emdebian-tools: [INTL:ru]
Russian debconf templates translation update</a>
Can debchange.pl not use SOAP for reliable data queries against the
BTS? (Otherwise this will happen again.)
(debchange.pl also reuses $bugs to refer to %bugs and a string $bugs from
`wget`.)
A temporary hack would be to use this regex:
if (/bugreport.cgi\?bug=([0-9]*).*?>\#[0-9]*: (.*)<\/a>/) {
and
690 $pkg = $PACKAGE if (not defined ($pkg));
to prevent an uninitialized variable warning.
--
Neil Williams
=============
http://www.data-freedom.org/
http://www.nosoftwarepatents.com/
http://www.linux.codehelp.co.uk/
pgp8NdKhB4tjs.pgp
Description: PGP signature
