April 2003 - Dillo-dev - lists.mailman3.com

Re: [Dillo-dev] Cookie locks?

by Jorge Arellano Cid

Lars, > > I understand unrestricted access to cookie jars can be bad for your health, > but it's annoying that you can't have two Dillos using cookies at the same > time. I usually have a Dillo running permanently, and when I get a URL in > my mail, I can click on it to open another Dillo (quickly!) -- but that > Dillo can't use cookies, which is frequently (e.g. in bugzilla) necessary. > Could we do a write-lock on it? Actually, we plan (with Jörgen) to make of cookies a dpi server. Cheers Jorge.-

April 25, 2003

Re: [Dillo-dev] Bookmarks

by Jorge Arellano Cid

On Tue, 22 Apr 2003, Lars Clausen wrote: > > Are there any plans to reintroduce a bookmarks menu? Any patches for it? Not really. Although dpis will have a way of setting menu items someday, it's very unlikely it will be enough to make a good implementation from it. It is possible to make a dpi that uses a plain GTK+ interface for bookmarks (think of a dpi as an external program that communicates with dillo using the dpi protocol), but this probably would be an overkill for your problem. > While the page looks nifty for the editing, I really don't like getting an > extra bookmark page into my history stack every time I want to go to a > bookmark. One way to solve this is to make dpi:/bm/ URIs "phantom URIs". That is, to flag them and tell the stack manager to get rid of them once they're no longer displayed. This needs some thought though, because you may have more than one dillo window. > In other news: Yay! Cut-and-paste! One of my main reasons to use other > browsers is now gone! Thanks! :) Cheers Jorge.-

April 25, 2003

Re: [Dillo-dev] Delayed release

by Chris Palmer

Andreas Schweitzer writes: > > We haven't got an accurate procedure to reproduce the crash > > (yet). This explanation may help: > > Here is an alternative method : simply RIGHT-click on the thumbnails > in said page : http://www.incagold.com/products/Speed/speedthief.php > > Curiously, on FreeBSD I "only" get a Gtk-WARNING **: > gtk_signal_emit_by_name(): could not find signal "link_pressed" in the > `DwTable' class ancestry a second attempt to right click the image > will pop up the menu. For what it's worth, on OpenBSD with Dillo from CVS about a week old, I get no warnings and no crashes when I try this test. -- Chris music is what numbers feel like San Francisco, CA http://www.nodewarrior.org/chris

April 25, 2003

selection and performance

by Thorben Thuermer

One of dillo's main advantages is it's incredible speed and efficiency, but somehow the selection code doesn't seem to keep up with this. it seems to have some scalability problem that makes it use excessive amounts of cpu when selecting large parts of complex pages (tables), and slows dillo down to a crawl even on my p3/750 (which dillo usually runs lightning fast on). to reproduce the problem, load: http://users.auriga.wearlab.de/~r00t/table.html (a table of 21x21 cells), and then try to select all cells by dragging from the top left to the lower right. cpu usage grows immensely, and near the end it slows down to a crawl. the script used to create the table (to avoid duplicate work if you need other test-cases): { echo '<html><body><table border=1>';for a in $(seq 0 20);do echo '<tr>'; for b in $(seq 0 20);do echo '<td>'"$a-$b"'</td>';done;echo '</tr>';done; echo '</table></body></html>';}>table.html if this is a known limitation, i'm sorry to bug you with it, but it seemed worth reporting. Greetings, Thorben Thuermer

April 24, 2003

[PATCH] Pth library support for threads

by Patrice Mandin

Hello dillo coders, I made a simple patch to add pth support for threads for Dillo. So platforms that do not have the pthread library can also use it. I am not very clever regarding thread programming, it may needs a bit more of work. I tested it on very simple pages and it worked for me. -- Patrice Mandin WWW: http://membres.lycos.fr/pmandin/ Programmeur Linux, Atari Spécialité: Développement, jeux

April 24, 2003

Re: [Dillo-dev] Delayed release

by Andreas Schweitzer

> Onto another topic: > > A last minute leak fix (iterators) left dillo a bit unstable. > We knew the iterators had a leak and it's great to have found it. > Now it will take some extra time to try to fix the problem and, > after that, stress-test the Copy&Paste code again. > > We haven't got an accurate procedure to reproduce the crash > (yet). This explanation may help: Here is an alternative method : simply RIGHT-click on the thumbnails in said page : http://www.incagold.com/products/Speed/speedthief.php Curiously, on FreeBSD I "only" get a Gtk-WARNING **: gtk_signal_emit_by_name(): could not find signal "link_pressed" in the `DwTable' class ancestry a second attempt to right click the image will pop up the menu. Cheers, Andreas -- **************************** NEW ADDRESS ****************************** Hamburger Sternwarte Universitaet Hamburg Gojenbergsweg 112 Tel. ++49 40 42891 4016 D-21029 Hamburg, Germany Fax. ++49 40 42891 4198

April 24, 2003

Delayed release

by Jorge Arellano Cid

Hi there, I just commited the code that cleanups the bookmarks server. It works like this: <q> /* * If this is our first run, kill the (old) bookmarks server, * and create the '~/.dillo/.072' file to flag what we did. */ static void Dillo_check_bm_srv12(char *dir) [...] </q> Yes, it uses the "[b]m_srv12" trick and some other precautions. Please test it: the whole point is that when installing the new version of dillo (072), the old bookmarks server may keep running and survive the updated. The new code detects the first dillo-072 run and issues a kill command to "bm_srv12". Onto another topic: A last minute leak fix (iterators) left dillo a bit unstable. We knew the iterators had a leak and it's great to have found it. Now it will take some extra time to try to fix the problem and, after that, stress-test the Copy&Paste code again. We haven't got an accurate procedure to reproduce the crash (yet). This explanation may help: <q> I think I have found a relatively new segfault bug : You need some sort of thumbtags like the screenshots on http://www.incagold.com/products/Speed/speedthief.php Now you have to move down the mouse and at the same time click on one image (i.e. click while still aiming for the image). If the button-push occures outside the image and the button-release inside the image, chances are high to get a segfault. This is a bit hard to reproduce, but I couldn't get an older version with selection to behave like that. It also happened a few times to me last night (after the last 'cvs up') but never before. </q> We'll be working on it. Cheers Jorge.-

April 24, 2003

Re: [Dillo-dev] Cookie locks?

by Gabriel Kihlman

Hi, Lars Clausen <lrclause(a)cs.uiuc.edu> writes: > I understand unrestricted access to cookie jars can be bad for your health, > but it's annoying that you can't have two Dillos using cookies at the same > time. I usually have a Dillo running permanently, and when I get a URL in > my mail, I can click on it to open another Dillo (quickly!) -- but that > Dillo can't use cookies, which is frequently (e.g. in bugzilla) necessary. > Could we do a write-lock on it? That is something I would appreciate as well. I have been running with this ugly patch that disables locking for atleast 6-7 months without a problem: --- src/cookies.c.orig Sun Jan 12 01:25:20 2003 +++ src/cookies.c Mon Feb 24 09:43:37 2003 @@ -31,7 +31,6 @@ #include "list.h" #include "cookies.h" -#define DEBUG_LEVEL 8 #include "debug.h" /* The maximum length of a line in the cookie file */ @@ -140,7 +139,7 @@ void a_Cookies_init() char line[LINE_MAXLEN]; /* Default setting */ - disabled = TRUE; + disabled = FALSE; /* Read and parse the cookie control file (cookiesrc) */ if (Cookie_control_init() != 0) { @@ -158,6 +157,7 @@ void a_Cookies_init() return; } +#if 0 /* Try to get a lock from the file descriptor */ disabled = (lockf(fileno(file_stream), F_TLOCK, 0) == -1); if (disabled) { @@ -166,7 +166,7 @@ void a_Cookies_init() fclose(file_stream); return; } - +#endif DEBUG_MSG(10, "Enabling cookies as from cookiesrc...\n"); cookies = g_hash_table_new(g_str_hash, g_str_equal);

April 24, 2003

Re: [Dillo-dev] Vera

by Nicola Girardi

> > Yes, I'm using it now. > > I can use it too, but it's not anti-aliased. xterm is able to use > anti-aliased TTF fonts (seemingly ONLY Luxi, but that's another issue), > so I know my X setup can use it. What do you have to do to get > anti-aliasing? Apply my patch, look in the archives. Should apply with minor tweaks. (Again, I hope this creeps in the official version.)

April 23, 2003

Re: [Dillo-dev] Vera works now

by Andreas Zwinkau

> Sounds like the "FontPath" settings in your XF86Config file do not > match the "catalogue" settings in your xfs config > (<X11etcdir>/fs/config). xfs is simply a decent cache, which can > provide fonts over the network(only need to install on one machine, > etc etc). Update your server config to use more font paths. You can > augment the list at run time with "xset fp+ /path/to/more/fonts" if > you don't want to touch XF86Config. > > One thing is to make sure that you run "xset fp rehash" after adding > new fonts and running "mkfontdir", since X doesn't automatically see > new fonts. I should have said that before, but only gave "xfstt > --sync". Sorry. I thought it was a xfs problem, so i did run it. But now i checked my X output and it says "Could not init font path element unix/:7100, removing from list!". So xfs is not used! I think the problem was mkfontdir. I thought fc-cache (gentoo) would do this, but it does not seem so. Xfs did mkfontdir or something similar on all directories and so Dillo can find them now. Thats my assumption. I checked .../fs/config and it is there. Thanks for help :D -- Andreas Zwinkau | web: andi.dasstellenwirinsinternet.de | mail: andi(a)buxach.de | jabber: beza1e1(a)amessage.de

April 23, 2003