Subject: HTTP::Cookies rejects cookies from host a.b for domain .a.b To: firstname.lastname@example.org Date: 11 Oct 2001 23:11:08 +0200 Organization: LRDE/EPITA http://www.lrde.epita.fr/ Hi people, I'm writing an HTTP client for SourceForge using LWP (libwww-perl 5.53). I need to login to SourceForge by POSTing a form, to get an identification cookie before further processing. The trouble is that the (queried) host 'sourceforge.net' sends me a cookie as follow: Set-Cookie: username=BLABLABLA; expires=Wed, 17-Oct-01 22:37:23 GMT; path=/; domain=.sourceforge.net and HTTP::Cookies reject it with the this debug output: HTTP::Cookies::extract_cookies: Domain .sourceforge.net does not match host sourceforge.net I have no idea who is right, it's the first time I work with cookies. (Reading RFC 2109 I'd vote for HTTP::Cookies, however other browsers seems to accept these cookies just fine.) I've modified my copy of Cookies.pm as follow to get around this.
Subject: Bundle::LWP problem (bad order) To: email@example.com Date: Fri, 21 Sep 2001 18:59:34 -0400 Hi Gisle, In Bundle::LWP, URI is first, and MIME::Base64 is third. But MIME::Base64 is in URI's PREREQ_PM, so that things get out of whack. The result is that in a Bundle I am working on (which has a ton of things in it), URI etc. doesn't get installed until the end, which throws other things off. Temporarily, I put MIME::Base64 before Bundle::LWP in my Bundle, but changing the order for your next release would probably be good. Thanks,
Subject: patch for libwww-perl-5.53 To: gisle@ActiveState.com Date: Sun, 14 Oct 2001 21:31:57 +0300 Organization: NetSoft srl Hi Gisle, The test 8 from robot/rules-dbm.t fails when AnyDBM_File uses GDBM_File. GDBM_File doesn't truncate the file when O_TRUNC is used because it doesn't know about this flag. Old data is kept in the database, and the test fails. An explicit clean-up of the old database makes the test pass. Thank you, Radu Greab
The old module is now LWP::Protocol::http10.
an 'Expect: 100-continue' header.
Patch contributed by Keiichiro Nagano <firstname.lastname@example.org>.
… ever. This patch probably slows down the module quite a bit. I really need to do the select() stuff by hand.