- Dillo-dev - lists.mailman3.com

[Dillo-dev]Cookies bugs in bugtrack #364, #370, #617 and others

by Diego Sáenz

searching cookies bugs in bugtrack y have found and try this 3 bugs Bug #364: Seems that it is corrected so the bug can be deleted. I want to tell that people need to be carefull when put a rule and write a url in the location bar: if one write the url without the www, a.org instead www.a.org for example, the page can load but will not match with a rule .a.org in cookiesrc. Another thing is that if the page send a cookie for www.a.org and you put in the location location bar a.org the cookie will not match and will not be send. Bug #370: I do not see a php register in that page ¿can anybody put another example? Bug #617: Cookies.dpi suffers this bug too. Another thing: In the patch i send to make cookies a dpi the line 38 of dillo/src/cookies.c says #define DEBUG_LEVEL 1 /*8*/ and must be #define DEBUG_LEVEL 8 One more thing: ¿have try anybody the cookies dpi patch? ¿what do he or she think? ¿have anybody found a bug? One curiosity: If you have a page that point with a link to another page loaded(yet) in the dillo cache and click the link to go to the page, the page loads in that way that if the 2 pages have the same decorations and are both at top seems that only content, not the decoration, of the page have change, but if you used the back and forward buttons you see how the page change. Example open a new dillo(you are in the splash page), go to a directory (file:/etc/), click in a directory of that directory(file:/etc/X11/),go back with the button and see how the page redraws, now click again in the same directory(file:/etc/X11/) and see how the page changes without seeing the redraw. It works in http pages while they are in the cache. It seems that it is because the a_Nav_push of the page happen in a gtk signal. I think this can be interesting to a possible dpi that managed the dillo file menus and buttons interface with a html page so updates can be transparent. I hoppe that it can work in the FLTK dillo too. Diego/DarkSpirit

April 28, 2005

Re: [Dillo-dev]"Medium priority" "Make cookies into a dpi server" task

by Diego Sáenz

Work done, almost the first version works and is ready to test. The patch works with current cvs. Please do an intensive test so bugs can be found fast. I am expecially interesting in the https cookies thing. A bad thing is that netscape cookies file format do not support ports for cookies, maybe with another file, maybe updating both files. Cookies are now managed out of dillo in the cookies.dpi and it uses a cookies file in netscape format so it can be used with other programs like for example wget. Between sessions the cookies get saved to ~/.dillo/cookies.txt, the old ~/.dillo/cookies is read too but not updated. Now cookies dpi have the state of the cookies and is the only that read and writes cookies.txt file. The diferent dillos running ask and send cookies to it. To minimize comunication between dpi and dillos all them read cookiesrc and only ask and send cookies for allowed sites. Cookies.dpi need to read cookiesrc so if a site is changed to deny cookies it deletes the cookies for that site in the cookies.txt file next time it load and write that file. All the work is implemented adding only three new dpi commands, two really, and a new send bloking dpi command function. When dillos wants the cookies for a site send the get_cookie dpi command to cookies dpi and this send the cookies in a get_cookies_answer dpi command. If an allowed site send a cookie dillos uses set_cookies dpi command to store the cookie in dpi. Use "dpidc stop" before changes to cookies.txt or cookies files or you can lost your changes. After cookies dpi reload the file all dillos will use the changes. If you changes the cookiesrc file use "dpidc stop" so cookies dpi reload the file. Only new opened dillos will use the changes. Diego/DarkSpirit

April 27, 2005

Re: [Dillo-dev]Using dillo on SourceForge

by Diego Sáenz

Very thanks. I will try it now. DarkSpirit/Diego El Tue, 26 Apr 2005 11:16:54 +0200 Jorge Arellano Cid <jcid(a)dillo.org> escribio: > Hi Diego, > > On Sun, Apr 24, 2005 at 09:18:40PM +0200, Jorge Arellano Cid wrote: > > On Sun, Apr 24, 2005 at 10:18:17AM +0200, Jorge Arellano Cid wrote: > > > Hi Diego, > > > > > > On Sat, Apr 23, 2005 at 02:34:07PM +0200, Diego Sáenz wrote: > > > > > > > > > > > > > Hmm, what's the status of the cookies dpi? > > > > It is almost done, but i need an answer to a big doubt before it. > > > > > > > > how can a function wait(with a timeout if posible) until the > > > > cookies dpi send the answer to dillo after dillo have send to dpi > > > > cookies a get_cookies command? > > > > > > Just let it block. Of course a non-blocking timeout or callback > > > approach can be used (as with DNS or HTTP), but in this case the > > > wait time is so short that it is not worth (it will be faster > > > than bookmarks). > > > > Unfortunately, after a long look into the code, I found out > > that there's no easy way to make it block (the natural way). > > > > Making it non-blocking is possible but non-trivial. I'll give > > it another try tomorrow... > > Here's the patch! (I will send to you, not to the list). > > After a long day of coding a really hairy hack to make a non- > blocking approach, It was almost working, but I didn't push it > further because it was too ugly! :-) > > The good part is that after all that thought, a natural way to > make it block came to me, so that's what the patch is. It has all > the necessary stuff to bind your work. > > It comes with a custom cookies.dpi server example, the query is > shown on stdout... Oh, it will be clear from the patch and its > comments... > > It was a heavy context switch, now I'm back to the FLTK port. > > -- > Cheers > Jorge.- > > _______________________________________________ > Dillo-dev mailing list > Dillo-dev(a)dillo.org > http://lists.auriga.wearlab.de/cgi-bin/mailman/listinfo/dillo-dev

April 26, 2005

Re: [Fwd: Re: [Dillo-dev]GTK 1.2 ..?]

by Jorge Arellano Cid

On Tue, Apr 05, 2005 at 05:06:52PM -0000, Ballance, Robert T wrote: > > > By the way, jcid@, I heard you were at FOSDEM, and were staying with > > > Mind.be, which one were you again, I was staying there with the > > > MirOS/MirBSD hackers, I stayed up all night working on ports, so I > > > probably saw you downstairs sometime hacking away (or listening to those > > > comical Plan9 fellows talk ;)) > > > > There were two guys with long black hair. One was the Plan9 > > speechman, the other was me. ...and, we spend more than an hour > > talking to each other that night. ;) > > > > BTW, there was a guy in the room, playin funny music on his > > notebook. Was it you? :-) > > Now I know who you were, heh, I was that blond young fellow, not the german > guy, I am/was the American. I had a shitty old IBM Thinkpad with a big BSD > sticker on the back, I came down later, after all the MirOS(/BSD) fellows had > gone to sleep. I seem to remember that laptop... Probably mine is older. > > That Plan9 fellow was awfully opinionated wasn't he? :-P Yes, but he also had a lot of knowledge, and he was willing to share and discuss it (a thing that's not very common), so I really enjoyed the talk. > > He discarded the FreeBSD Project as anything worth a damn on the grounds that > "how can a OS be respected that doesn't even have their own compiler?" Maybe it was an indirect pointer to Ken Thompson's reflections on trusting trust. > Btw, are you going to do anything GNU/specific with the FLTK port, or will > there be the possibility of porting Dillo/FLTK to Windows (blech! but still, > OSS on Windows is useful :)) AFAIK, there're still some Windows limitations as, for instance, not being possible to watch a pipe's activity with select(). We'll see... -- Cheers Jorge.-

April 26, 2005

Re: [Dillo-dev]Using dillo on SourceForge

by Jorge Arellano Cid

Hi Diego, On Sun, Apr 24, 2005 at 09:18:40PM +0200, Jorge Arellano Cid wrote: > On Sun, Apr 24, 2005 at 10:18:17AM +0200, Jorge Arellano Cid wrote: > > Hi Diego, > > > > On Sat, Apr 23, 2005 at 02:34:07PM +0200, Diego Sáenz wrote: > > > > > > > > > > Hmm, what's the status of the cookies dpi? > > > It is almost done, but i need an answer to a big doubt before it. > > > > > > how can a function wait(with a timeout if posible) until the > > > cookies dpi send the answer to dillo after dillo have send to dpi > > > cookies a get_cookies command? > > > > Just let it block. Of course a non-blocking timeout or callback > > approach can be used (as with DNS or HTTP), but in this case the > > wait time is so short that it is not worth (it will be faster > > than bookmarks). > > Unfortunately, after a long look into the code, I found out > that there's no easy way to make it block (the natural way). > > Making it non-blocking is possible but non-trivial. I'll give > it another try tomorrow... Here's the patch! (I will send to you, not to the list). After a long day of coding a really hairy hack to make a non- blocking approach, It was almost working, but I didn't push it further because it was too ugly! :-) The good part is that after all that thought, a natural way to make it block came to me, so that's what the patch is. It has all the necessary stuff to bind your work. It comes with a custom cookies.dpi server example, the query is shown on stdout... Oh, it will be clear from the patch and its comments... It was a heavy context switch, now I'm back to the FLTK port. -- Cheers Jorge.-

April 26, 2005

Re: [Dillo-dev]Using dillo on SourceForge

by Jorge Arellano Cid

On Sun, Apr 24, 2005 at 10:18:17AM +0200, Jorge Arellano Cid wrote: > Hi Diego, > > On Sat, Apr 23, 2005 at 02:34:07PM +0200, Diego Sáenz wrote: > > > > > > > Hmm, what's the status of the cookies dpi? > > It is almost done, but i need an answer to a big doubt before it. > > > > how can a function wait(with a timeout if posible) until the > > cookies dpi send the answer to dillo after dillo have send to dpi > > cookies a get_cookies command? > > Just let it block. Of course a non-blocking timeout or callback > approach can be used (as with DNS or HTTP), but in this case the > wait time is so short that it is not worth (it will be faster > than bookmarks). Unfortunately, after a long look into the code, I found out that there's no easy way to make it block (the natural way). Making it non-blocking is possible but non-trivial. I'll give it another try tomorrow... -- Cheers Jorge.-

April 24, 2005

Re: [Dillo-dev]Using dillo on SourceForge

by Jorge Arellano Cid

Hi Diego, On Sat, Apr 23, 2005 at 02:34:07PM +0200, Diego Sáenz wrote: > > > > Hmm, what's the status of the cookies dpi? > It is almost done, but i need an answer to a big doubt before it. > > how can a function wait(with a timeout if posible) until the > cookies dpi send the answer to dillo after dillo have send to dpi > cookies a get_cookies command? Just let it block. Of course a non-blocking timeout or callback approach can be used (as with DNS or HTTP), but in this case the wait time is so short that it is not worth (it will be faster than bookmarks). > more explained: > the a_Cookies_get function in dillo send a get_cookies command > to cookies dpi, then wait a command from the cookies dpi with the > answer and the function return the cookies. Yes, I understand. If you're curious just take a look at the chat command in bookmarks. There're three questions and answers for adding a bookmark! :-) > I have try to do the same that bookmarks dpi but the function > used in it is void. I have try to understand ccc but i get it > only superficialy. The CCC is hard to understand from the code (don't waste your time). I'll try to take a look into it today and send you some code that does the trick, so you can hook your work to it. -- Cheers Jorge.-

April 24, 2005

Re: [Dillo-dev]Using dillo on SourceForge

by Diego Sáenz

> > As far as I understand, the https dpi plugin does not handle cookies > > because the cookies do not find their way to cookies.c So you > > probably have to wait for the cookies dpi. can you give a url to test it? thanks > > Hmm, what's the status of the cookies dpi? It is almost done, but i need an answer to a big doubt before it. how can a function wait(with a timeout if posible) until the cookies dpi send the answer to dillo after dillo have send to dpi cookies a get_cookies command? more explained: the a_Cookies_get function in dillo send a get_cookies command to cookies dpi, then wait a command from the cookies dpi with the answer and the function return the cookies. I have try to do the same that bookmarks dpi but the function used in it is void. I have try to understand ccc but i get it only superficialy. DarkSpirit/Diego

April 23, 2005

Re: [Dillo-dev]Using dillo on SourceForge

by Jeremy Henty

On Sat, Apr 09, 2005 at 07:20:46PM +0200, Andreas Kemnade wrote: > On Mon, 4 Apr 2005 20:34:55 +0100 > Jeremy Henty <jeremy(a)chaos.org.uk> wrote: > > > Is the SSL-enabled version of dillo-0.8.4 up to managing > > SourceForge accounts? ... > > You need some of these ssl-patches (with an older dillo) which do > not use the dpi interface Are you saying I need to check out a previous version of dillo and forward-port some patches? If so, which version and which patches? > As far as I understand, the https dpi plugin does not handle cookies > because the cookies do not find their way to cookies.c So you > probably have to wait for the cookies dpi. Hmm, what's the status of the cookies dpi? Thanks for your help. Cheers, Jeremy

April 16, 2005

Re: [Dillo-dev]Using dillo on SourceForge

by Andreas Kemnade

On Mon, 4 Apr 2005 20:34:55 +0100 Jeremy Henty <jeremy(a)chaos.org.uk> wrote: > > Is the SSL-enabled version of dillo-0.8.4 up to managing SourceForge > accounts? I'm on the login page at > <URL:https://sourceforge.net/account/login.php> but when I click the > "Login with SSL" button the page just blinks at me as if I'd refreshed > the it. I don't get logged in. Is anyone using a SourceForge account > through dillo? > You need some of these ssl-patches (with an older dillo) which do not use the dpi interface because SF needs cookies to login. With that, you can use at least the SF installation at auriga.wearlab.de As far as I understand, the https dpi plugin does not handle cookies because the cookies do not find their way to cookies.c So you probably have to wait for the cookies dpi. Greetings Andreas Kemnade

April 9, 2005