From fonseca at diku.dk Mon Feb 4 16:54:30 2008 From: fonseca at diku.dk (Jonas Fonseca) Date: Tue, 5 Feb 2008 00:54:30 +0100 Subject: [elinks-users] [ANNOUNCE] ELinks 0.11.4rc0 Message-ID: <20080204235429.GA28006@diku.dk> Hello, Time is nearing another stable release, however, before hitting 0.11.4 a few release candidates seems to be in order to let any last pending changes migrate from the development branches. This release fixes a handful of critical bugs and security issues, including CVE-2007-2027, as well as some minor problems. You are adviced to upgrade! Below is the complete shortlog of changes since 0.11.3. Please see the NEWS file for a more structured list of fixes and improvements. Jonas Fonseca (9): elinks-0.11.3.GIT Never show empty filename in the what-to-do dialog Check if the program path contains "src/" before using ../po files Mention testing of po updates without installing ELinks Fix smjs scripting #ifdefs to use CONFIG_SM_SCRIPTING Update fuzzy messages in the Danish translation Change elinks.or.cz references to elinks.cz Use git tools instead of cogito for getting the build ID elinks-0.11.4rc0 Kalle Olavi Niemitalo (91): NEWS: List release dates, from mailing list announcements. NEWS: Open a section for 0.11.3.GIT. doc: More AsciiDoc compatibility. Updated Lua scripting document. lua_console_hook must return "goto_url", not "goto-url". lua-scripting: Revised examples for Lua 5.0. And other updates. Bug 945: Don't crash if the error message is not a string. NEWS: mention bug 945 NEWS: Clarify that these are bug numbers. l_pipe_read: Don't leak the old block if mem_realloc fails. AUTHORS: add Simon Josefsson Bug 712, ssl_set_no_tls: Disable TLS protocols for GnuTLS too. Bug 947, u2cp_: Use UCS_NO_BREAK_SPACE instead of 0xa0. Bug 879: New constant UCS_SOFT_HYPHEN; use where applicable. Bugs 879, 928, 947: Specially map U+00A0 and U+00AD in translation tables. Bug 947, set_hline: Respect wrap_nbsp also if !part->document. NEWS: mention bugs 928, 947 Document the meaning of document_options.cp. html_textarea: Use memmove rather than memcpy. Revert 2d6840b9, "Only set the socket protocol family on success." NEWS: mention passive FTP over IPv6 Debian bug 380347: Prevent a buffer overflow in entity_cache. NEWS: mention Debian bug 380347 NEWS: mention bug 396 NEWS: Use just http: instead of link:http:. NEWS: 0.11.3.GIT _now_ NEWS: mention bug 691 Bug 788: Define n_a variables if the POPpx macro of Perl needs them. NEWS: mention bug 788 configure.in: Advertise smb2.c to be included in ELinks 0.12.0. NEWS: bug 788 is more severe than mere warnings NEWS: mention CVE-2007-2027 Prepend $(top_builddir) to @INSTALL@ if it is relative. NEWS: mention config/install-sh on SunOS NEWS: markup changes from 0.12.GIT NEWS: Bug 950 now has a number. NEWS: mention the SpiderMonkey check build: Don't use $(AM_CFLAGS) anymore. Use $(CPPFLAGS) instead. NEWS: mention ${CPPFLAGS} vs. ${AM_CFLAGS} Bug 461: Ensure contrast when filling with spaces. NEWS: mention bug 461 List events in alphabetical order. List "Triggered When" before "Arguments" in the event template. Garbage-collect SMJS objects before flushing caches. NEWS: mention bug 951 NEWS: add severities of bugs and sort accordingly Revert 5a49b9f1, "NEWS: mention bug 951" Revert 644908c1, "Garbage-collect SMJS objects before flushing caches." NEWS: mention query parsing in file: URIs for local CGI Ignore src/ID and src/TAGS. Compare screen_driver.name with strcmp, not memcmp. NEWS: sync with 0.12.GIT NEWS: mention bug 956 (and bug 954) SMJS: Use JS_GetInstancePrivate where applicable. SMJS: comment changes Bug 755: A hack to make SMJS preserve integers as private data. NEWS: Bug 755 was fixed. Bug 957: Assert that the ECMAScript interpreter is not running. Bug 957: if_assert_failed in ecmascript_put_interpreter. Bug 960: Redefine LOADMSGCAT_USE_MMAP instead of HAVE_MMAP. NEWS: mention bug 960 Reapply 644908c1, "Garbage-collect SMJS objects before flushing caches." Reapply 5a49b9f1, "NEWS: mention bug 951" Link to gitweb at repo.or.cz, rather than pasky.or.cz. SMJS: Fix error "forms.namedItem is not a function". Bug 936: Create stamp-h.in in autogen.sh. Bug 936: Require Autoconf 2.61 or later. Bug 936: Allow Autoconf 2.59 too, for elinks.cz compatibility. NEWS: mention bug 936 NEWS: sync with 0.12.GIT Debian bug 431211: Correct GNU GPL version in debian/copyright. NEWS: mention bugs 613, 714, 961 NEWS: mention bug 959 configure.in: Actually the snapshots aren't daily. Remove gettext doc excerpt until the license is sorted out. Rewrite wiki to en.wikipedia.org, avoid server bug. NEWS: Bug 937 is CVE-2007-5034. Use -Wno-always-true only if $CC accepts it. GCC 4.1 doesn't. GCC 4.2.1 needs -Wno-address, not -Wno-always-true. NEWS: mention the builtin_modules warning Bug 975: Fix int/size_t pointer type mismatch. NEWS: Bug 975 was not in previous versions. Enable asciidoc7compatible mode. Document that GNU Make >= 3.78 is needed, and check it. Move debian to contrib/debian. NEWS: mention bug 989 get_pasv_socket: Use AF_INET6 when appropriate. Bug 997: Fix unlikely stack corruption in get_pasv_socket. NEWS: sort by severity Don't assume sizeof(int)==4 in bittorrent Detect all overflows in BitTorrent parse_bencoding_integer. Miciah Dashiel Butler Masters (5): Fix IPv4 DNS lookup bug Update SpiderMonkey configure check Debian compatibility Strings corrections from Malcolm Parsons Add KEYMAP_INVALID value for enum keymap_id and properly check for it Missing change from last commit Petr Baudis (2): Fix compilation of perl hooks Fix perl pre_format_html hook wrt. binary files Simon Josefsson (1): Use gnutls_set_default_priority. Witold Filipczyk (12): Fix query-part parsing in file: URIs (local CGI). Make SpiderMonkey consistent with SEE. Do not cache ecmascript_objs bug 755: Fixed. test, CGI: added a simple form to test CGI. contrib: Tool for view pages offline. check_html_form_hierarchy: Old code was buggy. configure.in: X11 - try link an executable using -lX11. Do not show always evaluate as true warnings. bug 978: Python's webbrowser.open_new_tab(URL) works since now. Polish translation was updated. The patch by Jakub Bogusz. pl.po: missing tilde Polish translation was updated. -- Jonas Fonseca From lars.bjorndal at broadpark.no Tue Feb 5 06:53:36 2008 From: lars.bjorndal at broadpark.no (Lars =?iso-8859-1?Q?Bj=F8rndal?=) Date: Tue, 05 Feb 2008 14:53:36 +0100 Subject: [elinks-users] Featurerequest Message-ID: Hello, list! Let me first tell you about a problem I often face. I'm blind, and I use elinks for most of my Internet surfing. It's convenient to use the search command to find a string on current web page. However, after searching, it's not easy for me to find the search result. After searching, the cursor is left back either on a link or at the bottom line of the screen, while the search result is displayed some other place on the screen. To find the string, I need to read through the entire or most of the screen content. So what about implement a sort of virtual link where the search result are displayed, that tracks the cursor onto that location? If that was possible, I would find the result immediately, as the braille display which I use, always displays the line which contains the cursor as the cursor moves. If such a virtual link was implemented into Elinks, I think also it could be fine to have commands to jump to next/previous heading, table, paragraph, text and so on. Sorry, but I'm not a programmer. It would be great if somebody here would like to implement such a feature, if people here think it would make sense. Best regards, Lars From steve at holmesgrown.com Thu Feb 7 05:00:26 2008 From: steve at holmesgrown.com (Steve Holmes) Date: Thu, 7 Feb 2008 05:00:26 -0700 Subject: [elinks-users] Featurerequest In-Reply-To: References: Message-ID: <20080207120026.GB24483@lnx3.holmesgrown.com> -----BEGIN PGP SIGNED MESSAGE----- Hash: RIPEMD160 I don't know if that will be very easy or not since the cursor only goes where thare are clickable links. It's needs like this and the continual increase of web sites that don't work with text browsers that lead me to using Gnome, Orca and Firefox. Actually, tht combo is working out greate for me. Just a thought. I believe it is the javascript limitations that keep these text browsers from working properly with so many sites. On Tue, Feb 05, 2008 at 02:53:36PM +0100, Lars Bj?rndal wrote: > Hello, list! > > Let me first tell you about a problem I often face. I'm blind, and I > use elinks for most of my Internet surfing. It's convenient to use the > search command to find a string on current web page. However, after > searching, it's not easy for me to find the search result. After > searching, the cursor is left back either on a link or at the bottom > line of the screen, while the search result is displayed some other > place on the screen. To find the string, I need to read through the > entire or most of the screen content. > > So what about implement a sort of virtual link where the search result > are displayed, that tracks the cursor onto that location? If that was > possible, I would find the result immediately, as the braille display > which I use, always displays the line which contains the cursor as the > cursor moves. > > If such a virtual link was implemented into Elinks, I think also it > could be fine to have commands to jump to next/previous heading, > table, paragraph, text and so on. > > Sorry, but I'm not a programmer. It would be great if somebody here > would like to implement such a feature, if people here think it would > make sense. > > Best regards, > Lars > _______________________________________________ > elinks-users mailing list > elinks-users at linuxfromscratch.org > http://linuxfromscratch.org/mailman/listinfo/elinks-users - -- HolmesGrown Solutions The best solutions for the best price! http://holmesgrown.ld.net/ -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.7 (GNU/Linux) iD8DBQFHqvLaWSjv55S0LfERA6+RAJwKh9m+3x/00ErSlgXqZdPVtiODKACffO3t /AFyN+MvL2lt1swNu0K7S9k= =jOLS -----END PGP SIGNATURE----- From kon at iki.fi Sat Feb 9 07:25:28 2008 From: kon at iki.fi (Kalle Olavi Niemitalo) Date: Sat, 09 Feb 2008 16:25:28 +0200 Subject: [elinks-users] Featurerequest In-Reply-To: References: Message-ID: <87odaqdx5z.fsf@Astalo.kon.iki.fi> lars.bjorndal at broadpark.no (Lars Bj?rndal) writes: > So what about implement a sort of virtual link where the search result > are displayed, that tracks the cursor onto that location? If that was > possible, I would find the result immediately, as the braille display > which I use, always displays the line which contains the cursor as the > cursor moves. ELinks already supports moving the cursor to places where there is no link: in the main keymap, you can bind e.g. move-cursor-down to a key. So the search code could just use that feature, instead of making a virtual link. It seems this would go somewhere in find_next_do, but I don't really understand the code, so no patch this time. From snk at gna.org Sun Feb 24 00:32:12 2008 From: snk at gna.org (Suraj N. Kurapati) Date: Sat, 23 Feb 2008 23:32:12 -0800 Subject: [elinks-users] dump html after javascript eval? Message-ID: <47C11D7C.9070405@gna.org> Hello, Is it possible to make elinks dump the HTML for a web page (from elinks' internal DOM representation of the web page) *after* it evaluates all embedded JavaScript inside the original HTML? For instance: [orig HTML + embedded JS] => elinks DOM => [pure HTML] This is pretty much like the dumping preformatted text feature, except we dump the pure HTML from elinks' internal DOM representation of the web page. I am asking this question because I have a bunch of HTML files that were purposely obfuscated (using JavaScript) by its creator. The HTML files contain embedded JavaScript which dynamically populates (using document.write()) the otherwise empty HTML page. I would like to effectively eval() the embedded JavaScript in-place and obtain the pure HTML. Thanks for your consideration. From onemda at gmail.com Tue Feb 26 21:08:15 2008 From: onemda at gmail.com (Paul B. Mahol) Date: Wed, 27 Feb 2008 05:08:15 +0100 Subject: [elinks-users] elinks locks on ftp symbolic link Message-ID: <3a142e750802262008l6fd55be5v44207bc4479dd3fc@mail.gmail.com> Summary: elinks locks on ftp symbolic link check for example ftp.freebsd.org/pub/FreeBSD/ISO-IMAGES-ia64/ ELinks 0.11.4rc0 (built on Feb 27 2008 04:58:12) Features: Standard, No Backtrace, IPv6, gzip, bzip2, Periodic Saving, Timer, Cascading Style Sheets, Protocol (File, FTP, HTTP, URI rewrite, User protocols), SSL (OpenSSL), MIME (Option system, Mailcap, Mimetypes files), LED indicators, Bookmarks, Cookies, Form History, Global History, Goto URL History, Search History From fonseca at diku.dk Wed Feb 27 05:29:43 2008 From: fonseca at diku.dk (Jonas Fonseca) Date: Wed, 27 Feb 2008 13:29:43 +0100 Subject: [elinks-users] elinks locks on ftp symbolic link In-Reply-To: <3a142e750802262008l6fd55be5v44207bc4479dd3fc@mail.gmail.com> References: <3a142e750802262008l6fd55be5v44207bc4479dd3fc@mail.gmail.com> Message-ID: <20080227122943.GA3093@diku.dk> Paul B. Mahol wrote Wed, Feb 27, 2008: > Summary: elinks locks on ftp symbolic link > > check for example ftp.freebsd.org/pub/FreeBSD/ISO-IMAGES-ia64/ Thank you for reporting and providing the example. I have commited a fix to the elinks-0.11 branch, so it will be in the next stable release. The patch is attached below in case you are interested. --- src/protocol/ftp/parse.c | 2 +- src/protocol/ftp/test-ftp-parser | 5 +++++ 2 files changed, 6 insertions(+), 1 deletions(-) diff --git a/src/protocol/ftp/parse.c b/src/protocol/ftp/parse.c index be4326e..8048406 100644 --- a/src/protocol/ftp/parse.c +++ b/src/protocol/ftp/parse.c @@ -345,7 +345,7 @@ parse_ftp_unix_response(struct ftp_file_info *info, unsigned char *src, int len) break; } - pos = memchr(pos, ' ', end - pos); + pos = memchr(pos + 1, ' ', end - pos); } if (!info->symlink.source) diff --git a/src/protocol/ftp/test-ftp-parser b/src/protocol/ftp/test-ftp-parser index 860056d..370e223 100644 --- a/src/protocol/ftp/test-ftp-parser +++ b/src/protocol/ftp/test-ftp-parser @@ -62,6 +62,11 @@ test_ftp_response_expect_success \ -rwxrwxrwx 1 noone nogroup 322 Aug 19 1996 message.ftp\r\n" test_ftp_response_expect_success \ +'Response from FreeBSD with symbolic link' \ +" +lrwxrwx--x 1 110 1002 13 Oct 24 2006 5.3 has moved to ftp-archive -> ../README.TXT\r\n" + +test_ftp_response_expect_success \ 'Response from NetWare' \ " d[R----F--] supervisor 512 Jan 16 18:53 login\r\n -- 1.5.4.2.1353.g94e17 -- Jonas Fonseca From fonseca at diku.dk Wed Feb 27 05:31:03 2008 From: fonseca at diku.dk (Jonas Fonseca) Date: Wed, 27 Feb 2008 13:31:03 +0100 Subject: [elinks-users] dump html after javascript eval? In-Reply-To: <47C11D7C.9070405@gna.org> References: <47C11D7C.9070405@gna.org> Message-ID: <20080227123103.GB3093@diku.dk> Suraj N. Kurapati wrote Sat, Feb 23, 2008: > Hello, > > Is it possible to make elinks dump the HTML for a web page (from > elinks' internal DOM representation of the web page) *after* it > evaluates all embedded JavaScript inside the original HTML? No, this is not possible. Sorry. -- Jonas Fonseca From b79net at gmail.com Thu Feb 28 11:19:09 2008 From: b79net at gmail.com (John Magolske) Date: Thu, 28 Feb 2008 10:19:09 -0800 Subject: [elinks-users] Ctrl-C -- possible to make it not quit ELinks? Message-ID: <20080228181909.GA3258@mail.b79.net> I picked up the habit of using Ctrl-C to close unresponsive connections while using w3m. Now that I'm using ELinks more & more, occasionally I make the mistake of hitting Ctrl-C, which will close an ELinks session along with all open tabs without warning. I'm re-training myself to use "z", which is currently mapped to "Abort connection". But would like to find a way to prevent accidentally closing ELinks in this way. Adding the "Ctrl-C" binding to "Abort connection" made no difference. In the list archives I found: > If you do e.g. "stty intr undef" so that Ctrl-C does not give > ELinks a signal, then ELinks will handle it as a bindable key. http://linuxfromscratch.org/pipermail/elinks-users/2006-December/001384.html But this disables the functionality of Ctrl-C altogether within the terminal. Are there any other options? Regards, John -- John Magolske http://B79.net/contact From fonseca at diku.dk Thu Feb 28 16:38:43 2008 From: fonseca at diku.dk (Jonas Fonseca) Date: Fri, 29 Feb 2008 00:38:43 +0100 Subject: [elinks-users] Ctrl-C -- possible to make it not quit ELinks? In-Reply-To: <20080228181909.GA3258@mail.b79.net> References: <20080228181909.GA3258@mail.b79.net> Message-ID: <20080228233843.GB12833@diku.dk> John Magolske wrote Thu, Feb 28, 2008: > I picked up the habit of using Ctrl-C to close unresponsive > connections while using w3m. Now that I'm using ELinks more & more, > occasionally I make the mistake of hitting Ctrl-C, which will close > an ELinks session along with all open tabs without warning. > > I'm re-training myself to use "z", which is currently mapped to > "Abort connection". But would like to find a way to prevent > accidentally closing ELinks in this way. > > Adding the "Ctrl-C" binding to "Abort connection" made no difference. Ctrl-C is sadly hardcoded deep down in the code and thus does not allow this keycombo to be overwritten. -- Jonas Fonseca