- Dillo-dev - lists.mailman3.com

background images and --local/URL_SpamSafe
by eocene＠gmx.com Sept. 23, 2014

Sept. 23, 2014

It looks like --local/URL_SpamSafe doesn't have any effect on background images.

2 1

Dillo and Irix
by csoren＠isd.net Sept. 19, 2014

Sept. 19, 2014

I stand corrected! 0_o Thanks guys... :) > > The gcc on my machine is the nekoware gcc 4.7.1 > compiled by shadeofblue. It is the standard gcc > for Irix currently. If __sgi is true it is really > a SGI system. > > > > > eocene wrote: > > > > > > > > Chris wrote: > > > #if defined(__sgi) > > > > > > This will work with the MIPSPro compiler but > > > I'm pretty sure the GNU compilers do not > > > #define __sgi > > > > > > > On Jan's machine, > > > > ./cpp --version > > cpp (GCC) 4.7.1 > > > > ./cpp -dM |grep __sgi > > #define __sgi 1 > > #define __sgi__ 1 > >

1 0

Dillo and Irix
by csoren＠isd.net Sept. 19, 2014

Sept. 19, 2014

> > > > #if defined(__sgi) > > This will work with the MIPSPro compiler but I'm pretty sure the GNU compilers do not #define __sgi

3 2

Problems tackling a bug
by sgeerken＠dillo.org Sept. 17, 2014

Sept. 17, 2014

Hi! I noticed that the current tip of dillo hangs in most cases when displaying <http://www.dillo.org/>. I succeeded to reproduce this bug in gdb, but get many "<optimized out>" values. So I tried to reduce optimization, but unfortunately had problems reproducing this bug: - "-O2" (standard): reproducable both stand-alone and in gdb; - "-O1": reproducable stand-alone, but not in gdb; - "-O0": not reproducable at all. Versions envolved: $ gcc --version gcc (Debian 4.7.2-5) 4.7.2 $ gdb --version GNU gdb (GDB) 7.4.1-debian Any ideas? Sebastian

2 2

Dillo and Irix
by csoren＠isd.net Sept. 16, 2014

Sept. 16, 2014

> > > > There might even be some builtin support for this check > > in autoconf as the issue is mentioned in [1]. > > > > I looked a bit yesterday and didn't get the impression > that there was anything. > There is an autoconf m4 macro for snprintf, with some IRIX specific checks, here: http://www.jhweiss.de/software/snprintf.html According to that page versions of IRIX all the way up to the present (6.5.30) don't adhere to the standard. For what it's worth this is the IRIX online manual page for snprintf(): http://techpubs.sgi.com/library/tpl/cgi-bin/getdoc.cgi?coll=0650&db=man&fna…

1 0

Browser tracking
by jcid＠dillo.org Sept. 16, 2014

Sept. 16, 2014

Hi, This site is interesting. It gives kind of a score of web browser trackability. [1] @corvid, I thought you'd like to give it a look (AFAIR, you have submitted some standardizing to our HTTP querying in the past). It looks like our HTTP_ACCEPT and "User Agent" are the most vulnerable to tracking. It looks simple to provide some less unique alternatives in dillorc. DISCLAIMER: this is just on little aspect of the privacy chain. Most probably this is moot unless you're behind a TOR ring. YMMV. [1] https://panopticlick.eff.org/ -- Cheers Jorge.-

5 6

Patch: suppress duplicate Fl_Printer::class_id
by james.from.wellington＠gmail.com Sept. 16, 2014

Sept. 16, 2014

Hello All, Building commit 317f010 (two days ago) of dillo, I got a duplicate symbol on OSX 10.6.8 with gcc 4.2.1. Here is the diff that suppresses it. ---------------- $ hg diff diff -r c09473fb7311 dw/fltkplatform.cc --- a/dw/fltkplatform.cc Sat Sep 13 09:46:28 2014 +0000 +++ b/dw/fltkplatform.cc Sat Sep 13 23:49:26 2014 +1200 @@ -37,10 +37,10 @@ * any need for currently. This stub prevents that. */ class FL_EXPORT Fl_Printer : public Fl_Paged_Device { public: - static const char *class_id; +// static const char *class_id; Fl_Printer(void) {}; }; -const char *Fl_Printer::class_id = "Fl_Printer"; +//const char *Fl_Printer::class_id = "Fl_Printer"; /* Tooltips */ static Fl_Menu_Window *tt_window = NULL; --------------- I have no idea what it will turn out that I've broken. If having this as a pull request will help, then I will try to work out how. Regards, James.

2 2

dillo crash when dpid missing and view source used
by eocene＠gmx.com Sept. 16, 2014

Sept. 16, 2014

...it happens at least as far back as 3.0.4 release.

2 1

Implementing </div> closing <span>, and friends
by james.from.wellington＠gmail.com Sept. 15, 2014

Sept. 15, 2014

Hello all, [Sorry if this is a dup; gmail may be playing tricks on me.] This page displays, with distortions, in dillo 3.0.4, but the image does not display after commit 317f010: https://medium.com/@verynicetweets/getting-out-of-the-house-is-so-hard-thes… A partially cleaned-up and stripped version is attached, where the comment block shows a suppression of the problem. The page is illegal: two unclosed <span>s interfere with a </div> and the following <div>. There is code around html.cc:1355, which is intended to implement heuristics to close unclosed tags but, in the cases where it is triggered, no tags are closed. Here is a patch which I think implements the intent of the heuristics: ----------------- $ hg diff diff -r 26378f85c4ea src/html.cc --- a/src/html.cc Sun Sep 14 09:56:22 2014 +0000 +++ b/src/html.cc Mon Sep 15 18:13:21 2014 +1200 @@ -1355,7 +1355,9 @@ i_SELECT = a_Html_tag_index("select"), i_TEXTAREA = a_Html_tag_index("textarea"); int w3c_mode = !prefs.w3c_plus_heuristics; - int stack_idx, tag_idx, matched = 0, expected = 0; + // TODO: If there was a clean way of never having a tag index zero, + // (it's currently <a>), then expected_idx could be prettier. + int stack_idx, tag_idx, matched = 0, expected_idx = -1, skipped = 0; TagInfo new_tag = Tags[new_idx]; /* Look for the candidate tag to close */ @@ -1368,26 +1370,42 @@ break; } else if (Tags[tag_idx].EndTag == 'O') { /* skip an optional tag */ + ++skipped; continue; } else if ((new_idx == i_BUTTON && html->InFlags & IN_BUTTON) || (new_idx == i_SELECT && html->InFlags & IN_SELECT) || (new_idx == i_TEXTAREA && html->InFlags & IN_TEXTAREA)) { /* let these elements close tags inside them */ + ++skipped; continue; } else if (w3c_mode || Tags[tag_idx].TagLevel >= new_tag.TagLevel) { /* this is the tag that should have been closed */ - expected = 1; - break; + /* skip it as if it were optional, but complain */ + ++skipped; + if (-1 == expected_idx) { + // ie this is the first thing we're unwinding + expected_idx = tag_idx; + } + continue; } } + // Either or both matched and expected_idx could now be set + + // TODO: decide whether to carry line numbers in the parser state stack, + // in order to report them here. + if (matched) { + if (-1 != expected_idx) { + BUG_MSG("Unexpected closing tag: </%s> -- expected </%s>", + new_tag.name, Tags[expected_idx].name); + } + Html_tag_cleanup_to_idx(html, stack_idx); - } else if (expected) { - BUG_MSG("Unexpected closing tag: </%s> -- expected </%s>.", - new_tag.name, Tags[tag_idx].name); } else { - BUG_MSG("Unexpected closing tag: </%s>.", new_tag.name); + BUG_MSG("Unexpected closing tag: </%s>" + " -- explored %d layer(s) and not closed.", + new_tag.name, skipped); } } ----------------- Closing tags that match no opening tag (I used </area>, which is illegal) do not cause catastrophic failure. So far, it has not broken on my usual workload. Gmail is noisy, so I wonder if something's going wrong there, but it looks fairly normal. I have no good idea what a general test suite for this code looks like. If someone wants to point me toward related tests, then I will think about extending them. Regards, James.

1 0

dillo can't handle strings
by mail＠gabucino.hu Sept. 14, 2014

Sept. 14, 2014

See the issue on IRIX: http://lists.dillo.org/pipermail/dillo-dev/ 2014-April/010079.html Well, the same applies to PowerPC (OS X). "ftp://" works, "http://" doesn't. Somewhere there's a huge bug. Does anybody care? -- G

3 7