From mashdot at toshine.net Thu Oct 1 02:14:29 2009 From: mashdot at toshine.net ('Mash) Date: Thu, 1 Oct 2009 10:14:29 +0100 Subject: [Conkeror] Google results filter... In-Reply-To: <1254324313.81659.211.camel@Ares.God> References: <20090930101138.GA19986@grace.toshine.net> <20090930143455.GA16358@hecubus> <1254324313.81659.211.camel@Ares.God> Message-ID: <20091001091429.GF25812@grace.toshine.net> On 2009-09-30 17:25+0200, Ricardo Buring wrote: > On Wed, 2009-09-30 at 10:34 -0400, John J. Foerch wrote: > > On Wed, Sep 30, 2009 at 11:11:38AM +0100, 'Mash wrote: > > > I just had an idea... > > > > > > Would it be possible to create an optional "g" (Google search) to use a > > > defined list of sites you wish to filter out of the search, using > > > Google's "-term" ? > > > > > > For instance in your .rc file you could create a list of sites or terms > > > > > > filterList = ('wikipedia','bigresource.com') > > > > > > and then say using "gf" you would get: > > > > > > http://www.google.com/search?q=javascript+function+-wikipedia+-bigresource.com > > > > > > > > > I have no idea how to implement this even in a webjump without spending > > > hours; but one of you lot could surely whip-up a webjump or expand the > > > Google module. > > > > > > The idea came when I wanted to create a easy way to filter out the > > > horror that is "bigresource.com" from my results. > > > > > > Doesn't google itself have this feature? Next to each search result, > > there is an [X] button. Not 100% sure. > > > > To do this in conkeror would entail adding a feature to > > google-search-results-mode, not creating a webjump. > > > > I haven't looked into Google modes, but you can put this quick > Javascript in your .conkerorrc for a webjump: > > google_blacklist_array = ["wikipedia", "bigresource.com"]; > google_blacklist = ""; > for(var i=0; i { > google_blacklist += "+-" + google_blacklist_array[i]; > } > define_webjump("gf", "http://www.google.com/search?q=%s"+google_blacklist); > > That's probably not the best way to do it, but it works :) Thanks! That will do nicely. 'Mash --- Sola fide, Sola Gratia, Soli Deo gloria! From jjfoerch at earthlink.net Sat Oct 3 11:29:06 2009 From: jjfoerch at earthlink.net (John J. Foerch) Date: Sat, 3 Oct 2009 14:29:06 -0400 Subject: [Conkeror] github Message-ID: <20091003182906.GA31109@hecubus> Hi all, In response to a request for this, I made a conkeror repository on github. This is a secondary repository. The one on repo.or.cz is still our main repository, but I will make a point to keep master branch up to date in this one as well. The url is: http://github.com/retroj/conkeror Github is a useful tool for anyone wishing to have their own fork of conkeror from which we can easily pull patches. -- John Foerch From jjfoerch at earthlink.net Sat Oct 3 11:51:31 2009 From: jjfoerch at earthlink.net (John J. Foerch) Date: Sat, 3 Oct 2009 14:51:31 -0400 Subject: [Conkeror] new-input-system Message-ID: <20091003185131.GB31109@hecubus> Hi all, Yesterday I pushed a branch called new-input-system. It contains a patch which is a substantial rewrite of conkeror's event handling system. It resolves the keydown/keyup problems we have with some websites (yahoo, slashdot, ...) as well as laying in place the basis for better mouse support and more powerful browser objects. I don't expect any major problems merging this into master, but because of the size of the patch, I want to have it available for a short period of testing by the community before merge. If you are running conkeror from our main git repository, you can help test this patch. Just do the following: git fetch git checkout -b new-input-system origin/new-input-system When you want to go back to master branch, just do: git checkout master If you find any problems, you can post to the list, email me, or better yet, talk to me on our irc channel. Thanks! -- John Foerch From eichin at gmail.com Sat Oct 3 14:11:01 2009 From: eichin at gmail.com (Mark Eichin) Date: Sat, 3 Oct 2009 17:11:01 -0400 Subject: [Conkeror] using loadsubscript Message-ID: What I'm really trying to do is load jquery into an existing page, and then load my own code, so I can prototype/demo some improvements to a website I can't otherwise modify. interactive("jquery-test", "load and run a jquery file", function(I) { subscript_loader.loadSubScript("file://" + "/usr/share/javascript/jquery/jquery.js", I.buffer.document); subscript_loader.loadSubScript("file://" + "/tmp/test.js", I.buffer.document); }); This definitely loads and runs jquery.js and test.js; if I put alert() calls in test.js they fire, but $() references in test.js fail with "ReferenceError: $ is not defined." Actually, now that I look more closely jquery isn't just logging stylistic warnings, it's also reporting Console error: [JavaScript Warning: "reference to undefined property jQuery.cache[id][name]" {file: "chrome://conkeror-modules/content/rc.js -> file:///home/eichin/.emacs.js -> file:///usr/share/javascript/jquery/jquery.js" line: 679}] Category: component javascript Any suggestions? Is I.buffer.document not the right target object? -- _Mark_ From jjfoerch at earthlink.net Sat Oct 3 15:23:50 2009 From: jjfoerch at earthlink.net (John J. Foerch) Date: Sat, 3 Oct 2009 18:23:50 -0400 Subject: [Conkeror] using loadsubscript In-Reply-To: References: Message-ID: <20091003222350.GA2864@hecubus> On Sat, Oct 03, 2009 at 05:11:01PM -0400, Mark Eichin wrote: > What I'm really trying to do is load jquery into an existing page, and > then load my own code, so I can prototype/demo some improvements to a > website I can't otherwise modify. > > interactive("jquery-test", > "load and run a jquery file", > function(I) { > subscript_loader.loadSubScript("file://" + > "/usr/share/javascript/jquery/jquery.js", I.buffer.document); > subscript_loader.loadSubScript("file://" + "/tmp/test.js", I.buffer.document); > }); > > This definitely loads and runs jquery.js and test.js; if I put alert() > calls in test.js they fire, but $() references in test.js fail with > "ReferenceError: $ is not defined." Actually, now that I look more > closely jquery isn't just logging stylistic warnings, it's also > reporting > Console error: [JavaScript Warning: "reference to undefined property > jQuery.cache[id][name]" > {file: "chrome://conkeror-modules/content/rc.js -> > file:///home/eichin/.emacs.js -> > file:///usr/share/javascript/jquery/jquery.js" > line: 679}] > Category: component javascript > > Any suggestions? Is I.buffer.document not the right target object? The method that I'm familiar with involves using the DOM to add html:script elements to the document's head to load your js files. -- John Foerch From justloop at gmail.com Sat Oct 3 15:29:07 2009 From: justloop at gmail.com (justloop) Date: Sat, 3 Oct 2009 22:29:07 +0000 (UTC) Subject: [Conkeror] github References: <20091003182906.GA31109@hecubus> Message-ID: John J. Foerch earthlink.net> writes: > > Hi all, > > In response to a request for this, I made a conkeror repository on > github. This is a secondary repository. The one on repo.or.cz is still > our main repository, but I will make a point to keep master branch up to > date in this one as well. The url is: > > http://github.com/retroj/conkeror > > Github is a useful tool for anyone wishing to have their own fork of > conkeror from which we can easily pull patches. > I might be missing something but I get: yavuz at igor ~/src $ git clone http://github.com/retroj/conkeror Initialized empty Git repository in /Users/yavuz/src/conkeror/.git/ fatal: http://github.com/retroj/conkeror/info/refs not found: did you run git update-server-info on the server? From justloop at gmail.com Sat Oct 3 15:35:53 2009 From: justloop at gmail.com (justloop) Date: Sat, 3 Oct 2009 22:35:53 +0000 (UTC) Subject: [Conkeror] github References: <20091003182906.GA31109@hecubus> Message-ID: justloop gmail.com> writes: Silly me, of course it worked after I changed the 'http:' to 'git:'. Sorry. From eichin at gmail.com Sat Oct 3 22:31:51 2009 From: eichin at gmail.com (Mark Eichin) Date: Sun, 4 Oct 2009 01:31:51 -0400 Subject: [Conkeror] using loadsubscript In-Reply-To: <20091003222350.GA2864@hecubus> References: <20091003222350.GA2864@hecubus> Message-ID: Oh, that makes more sense. Don't even need to do anything special to make it activate - just M-x jquery-here RET $("h1").css("backgroundColor", "blue"); RET and the headings turn blue :-) function jquery_this_doc(d, js_code) { var script_el = d.createElementNS(XHTML_NS, "script"); script_el.setAttribute("language", "javascript"); script_el.setAttribute("type", "text/javascript"); script_el.setAttribute("src", "http://ajax.googleapis.com/ajax/libs/jquery/1.3.2/jquery.min.js"); d.body.appendChild(script_el); script_el = d.createElementNS(XHTML_NS, "script"); script_el.textContent = js_code; d.body.appendChild(script_el); } interactive("jquery-here", "load jquery.js into this page, then your command", function(I) { jquery_this_doc(I.buffer.document, (yield I.minibuffer.read($prompt = "jq: ", $history = "jquery-here"))); }); From jjfoerch at earthlink.net Sun Oct 4 10:08:34 2009 From: jjfoerch at earthlink.net (John J. Foerch) Date: Sun, 4 Oct 2009 13:08:34 -0400 Subject: [Conkeror] using loadsubscript In-Reply-To: References: <20091003222350.GA2864@hecubus> Message-ID: <20091004170834.GA10235@hecubus> On Sun, Oct 04, 2009 at 01:31:51AM -0400, Mark Eichin wrote: > Oh, that makes more sense. Don't even need to do anything special to > make it activate - just M-x jquery-here RET > $("h1").css("backgroundColor", "blue"); RET and the headings turn blue > :-) > > function jquery_this_doc(d, js_code) { > var script_el = d.createElementNS(XHTML_NS, "script"); > script_el.setAttribute("language", "javascript"); > script_el.setAttribute("type", "text/javascript"); > script_el.setAttribute("src", > "http://ajax.googleapis.com/ajax/libs/jquery/1.3.2/jquery.min.js"); > d.body.appendChild(script_el); > script_el = d.createElementNS(XHTML_NS, "script"); > script_el.textContent = js_code; > d.body.appendChild(script_el); > } > > interactive("jquery-here", > "load jquery.js into this page, then your command", > function(I) { > jquery_this_doc(I.buffer.document, (yield I.minibuffer.read($prompt > = "jq: ", $history = "jquery-here"))); > }); So it works now? Cool.. maybe put it on the wiki. Maybe a new page about doing web development with conkeror. -- John Foerch From jjfoerch at earthlink.net Mon Oct 5 08:29:45 2009 From: jjfoerch at earthlink.net (John J. Foerch) Date: Mon, 5 Oct 2009 11:29:45 -0400 Subject: [Conkeror] new-input-system In-Reply-To: <20091003185131.GB31109@hecubus> References: <20091003185131.GB31109@hecubus> Message-ID: <20091005152945.GA21042@hecubus> On Sat, Oct 03, 2009 at 02:51:31PM -0400, John J. Foerch wrote: > Yesterday I pushed a branch called new-input-system. It contains a > patch which is a substantial rewrite of conkeror's event handling system. > It resolves the keydown/keyup problems we have with some websites (yahoo, > slashdot, ...) as well as laying in place the basis for better mouse > support and more powerful browser objects. > > I don't expect any major problems merging this into master, but because > of the size of the patch, I want to have it available for a short period > of testing by the community before merge. If you are running conkeror > from our main git repository, you can help test this patch. Just do the > following: > > Anyone testing this branch, please update. -- John Foerch From jjfoerch at earthlink.net Mon Oct 5 09:27:09 2009 From: jjfoerch at earthlink.net (John J. Foerch) Date: Mon, 5 Oct 2009 12:27:09 -0400 Subject: [Conkeror] Can I change the font for a specific site? In-Reply-To: <83skebd8go.fsf@ymail.com> References: <83skebd8go.fsf@ymail.com> Message-ID: <20091005162709.GB21042@hecubus> On Fri, Sep 25, 2009 at 11:11:35AM +0800, Water Lin wrote: > > While I am exploring some sites in my Conkeror, the font size is too > small and makes my eyes very tired. Is there a way to set the font size > of this specific site? > > Thanks Hi Water, You can do this with css. Here is an example from my .conkerorrc, in which I use css to change the appearance of a particular website: // make the nasa-hsf-sightings page more printer friendly register_user_stylesheet( "data:text/css," + escape ( "@-moz-document url-prefix("+ "http://spaceflight1.nasa.gov/realdata/sightings/cities/view.cgi)"+ "{img { display: none; }}")); Note, I used a data: url to encode the css directly into my rc, which is convenient for short snippets like this. However, you can also put your css in a file, and pass a file: url to register_user_stylesheet. You may also need to mark your css rules as !important, depending on the particulars of the website's stylesheets. -- John Foerch From jackhill at MIT.EDU Mon Oct 5 13:53:47 2009 From: jackhill at MIT.EDU (Jack Hill) Date: Mon, 5 Oct 2009 16:53:47 -0400 (EDT) Subject: [Conkeror] new-input-system In-Reply-To: <20091005152945.GA21042@hecubus> References: <20091003185131.GB31109@hecubus> <20091005152945.GA21042@hecubus> Message-ID: xI have been using it for my daily browsing needs since you pushed it. I haven't noticed any bugs, but I also didn't notice the problems this addressed. Jack On Mon, 5 Oct 2009, John J. Foerch wrote: > On Sat, Oct 03, 2009 at 02:51:31PM -0400, John J. Foerch wrote: >> Yesterday I pushed a branch called new-input-system. It contains a >> patch which is a substantial rewrite of conkeror's event handling system. >> It resolves the keydown/keyup problems we have with some websites (yahoo, >> slashdot, ...) as well as laying in place the basis for better mouse >> support and more powerful browser objects. >> >> I don't expect any major problems merging this into master, but because >> of the size of the patch, I want to have it available for a short period >> of testing by the community before merge. If you are running conkeror >> from our main git repository, you can help test this patch. Just do the >> following: >> >> > > > Anyone testing this branch, please update. > > -- > John Foerch > _______________________________________________ > Conkeror mailing list > Conkeror at mozdev.org > https://www.mozdev.org/mailman/listinfo/conkeror > From jjfoerch at earthlink.net Tue Oct 6 12:46:15 2009 From: jjfoerch at earthlink.net (John J. Foerch) Date: Tue, 6 Oct 2009 15:46:15 -0400 Subject: [Conkeror] new-input-system In-Reply-To: <20091005152945.GA21042@hecubus> References: <20091003185131.GB31109@hecubus> <20091005152945.GA21042@hecubus> Message-ID: <20091006194615.GA6261@hecubus> On Mon, Oct 05, 2009 at 11:29:45AM -0400, John J. Foerch wrote: > On Sat, Oct 03, 2009 at 02:51:31PM -0400, John J. Foerch wrote: > > Yesterday I pushed a branch called new-input-system. It contains a > > patch which is a substantial rewrite of conkeror's event handling system. > > It resolves the keydown/keyup problems we have with some websites (yahoo, > > slashdot, ...) as well as laying in place the basis for better mouse > > support and more powerful browser objects. > > > > I don't expect any major problems merging this into master, but because > > of the size of the patch, I want to have it available for a short period > > of testing by the community before merge. If you are running conkeror > > from our main git repository, you can help test this patch. Just do the > > following: > > > > > > > Anyone testing this branch, please update. And again please. Probably about ready to merge. -- John Foerch From aditya.siram at gmail.com Wed Oct 7 07:04:32 2009 From: aditya.siram at gmail.com (aditya siram) Date: Wed, 7 Oct 2009 09:04:32 -0500 Subject: [Conkeror] Name Suggestion Message-ID: <594f78210910070704hb6a4ff0tcbcb167c73bd61f3@mail.gmail.com> NeTraps / Nettraps/ NetTraps - (trapping the net, as with a hair net, it is approximately Spartan spelled backward) -deech -------------- next part -------------- An HTML attachment was scrubbed... URL: From jjfoerch at earthlink.net Wed Oct 7 08:13:57 2009 From: jjfoerch at earthlink.net (John J. Foerch) Date: Wed, 7 Oct 2009 11:13:57 -0400 Subject: [Conkeror] Name Suggestion In-Reply-To: <594f78210910070704hb6a4ff0tcbcb167c73bd61f3@mail.gmail.com> References: <594f78210910070704hb6a4ff0tcbcb167c73bd61f3@mail.gmail.com> Message-ID: <20091007151357.GA13286@hecubus> On Wed, Oct 07, 2009 at 09:04:32AM -0500, aditya siram wrote: > NeTraps / Nettraps/ NetTraps - (trapping the net, as with a hair net, it is > approximately Spartan spelled backward) > > -deech Haha, you've won an "lol". This project is going to keep the name Conkeror, while another one, proposed as a Conkeror-like browser written in Scheme for Webkit, is to be named Minno. Problem solved. ;) -- John Foerch From jjfoerch at earthlink.net Thu Oct 8 11:24:41 2009 From: jjfoerch at earthlink.net (John J. Foerch) Date: Thu, 8 Oct 2009 14:24:41 -0400 Subject: [Conkeror] new-input-system In-Reply-To: <20091003185131.GB31109@hecubus> References: <20091003185131.GB31109@hecubus> Message-ID: <20091008182441.GA3848@hecubus> On Sat, Oct 03, 2009 at 02:51:31PM -0400, John J. Foerch wrote: > Hi all, > > Yesterday I pushed a branch called new-input-system. It contains a > patch which is a substantial rewrite of conkeror's event handling system. > It resolves the keydown/keyup problems we have with some websites (yahoo, > slashdot, ...) as well as laying in place the basis for better mouse > support and more powerful browser objects. To everyone who helped test this, thank you. I have merged it into master. To switch back to the master branch, do: git checkout master Thanks! -- John Foerch From dcl441-bugs at yahoo.com Fri Oct 9 00:21:14 2009 From: dcl441-bugs at yahoo.com (Daniel Clemente) Date: Fri, 09 Oct 2009 09:21:14 +0200 Subject: [Conkeror] corrections for transpose-chars References: <87my4kwnds.fsf@yahoo.com> <7b501d5c0909240753rc5c8333q673501f6a1e8415@mail.gmail.com> <87my4fzih2.fsf@yahoo.com> <20090928155121.GA23782@hecubus> Message-ID: <87d44xowwl.fsf@yahoo.com> El dl, set 28 2009 a les 17:51, John J. Foerch va escriure: > Grep for define_builtin_commands.. incidentally this needs to be > rewritten, but since it works "good enough" it's not at the top of my > to-do. Could you check this patch in? -------------- next part -------------- A non-text attachment was scrubbed... Name: transpose.diff Type: text/x-diff Size: 1315 bytes Desc: not available URL: -------------- next part -------------- Thanks From jjfoerch at earthlink.net Fri Oct 9 11:51:59 2009 From: jjfoerch at earthlink.net (John J. Foerch) Date: Fri, 9 Oct 2009 14:51:59 -0400 Subject: [Conkeror] corrections for transpose-chars In-Reply-To: <87d44xowwl.fsf@yahoo.com> References: <87my4kwnds.fsf@yahoo.com> <7b501d5c0909240753rc5c8333q673501f6a1e8415@mail.gmail.com> <87my4fzih2.fsf@yahoo.com> <20090928155121.GA23782@hecubus> <87d44xowwl.fsf@yahoo.com> Message-ID: <20091009185159.GA17254@hecubus> On Fri, Oct 09, 2009 at 09:21:14AM +0200, Daniel Clemente wrote: > El dl, set 28 2009 a les 17:51, John J. Foerch va escriure: > > Grep for define_builtin_commands.. incidentally this needs to be > > rewritten, but since it works "good enough" it's not at the top of my > > to-do. > > Could you check this patch in? > Thanks. pushed. -- John Foerch From jjfoerch at earthlink.net Mon Oct 12 16:49:59 2009 From: jjfoerch at earthlink.net (John J. Foerch) Date: Mon, 12 Oct 2009 19:49:59 -0400 Subject: [Conkeror] breaking change: mime_type_external_handlers Message-ID: <20091012234959.GA21774@hecubus> Hi all, I pushed a rewrite of the mime-type-external-handlers API, so if you have configured viewers for certain mime types in your rc, you will want to take note. The variable mime_type_external_handlers no longer exists, and it has been replaced by one called external_content_handlers. Here is how to use it, in short: external_content_handlers.set("application/pdf", "xpdf"); external_content_handlers.set("image/*", "display"); The way to "unset" a handler for a mime-type is to give null as the value. You can also set all handlers at once (overriding all defaults): external_content_handlers = { "*": "kate", //no idea why anybody would use kate, but anyway.. text: { "*": "kate" }, image: { "*": "feh" }, video: { "*": "gmplayer" }, audio: { "*": "gmplayer" }, application: { pdf: "xpdf", postscript: "gv", "x-dvi": "xdvi" } }; The main feature of this new patch was actually something called content_handlers. With the addition of that feature, it made good sense to overhaul the mime-type-external-handlers stuff along with it. I would encourage everyone to have a look at what the new feature does at: http://conkeror.org/ContentHandlers You just might find it convenient. -- John Foerch From dcl441-bugs at yahoo.com Tue Oct 13 08:59:49 2009 From: dcl441-bugs at yahoo.com (Daniel Clemente) Date: Tue, 13 Oct 2009 17:59:49 +0200 Subject: [Conkeror] breaking change: mime_type_external_handlers References: <20091012234959.GA21774@hecubus> Message-ID: <87ocobuvwq.fsf@yahoo.com> Thanks, that is simple and useful. >From now I can open PDF files directly with evince with this configuration: content_handlers.set("application/pdf", content_handler_open_default_viewer); external_content_handlers.set("application/pdf", "evince"); -- Daniel From jitenvdedhia at gmail.com Thu Oct 29 00:47:41 2009 From: jitenvdedhia at gmail.com (Jiten Dedhia) Date: Thu, 29 Oct 2009 00:47:41 -0700 (PDT) Subject: [Conkeror] Invitation to connect on LinkedIn Message-ID: <2127205505.99923.1256802461144.JavaMail.app@ech3-cdn06.prod> LinkedIn ------------ I'd like to add you to my professional network on LinkedIn. - Jiten Confirm that you know Jiten Dedhia https://www.linkedin.com/e/isd/827772504/aq7KTuLh/ Every day, millions of professionals like Jiten Dedhia use LinkedIn to connect with colleagues, find experts, and explore opportunities. ------ (c) 2009, LinkedIn Corporation -------------- next part -------------- An HTML attachment was scrubbed... URL: From jjfoerch at earthlink.net Thu Oct 29 11:12:49 2009 From: jjfoerch at earthlink.net (John J. Foerch) Date: Thu, 29 Oct 2009 14:12:49 -0400 Subject: [Conkeror] hints_auto_exit_delay default Message-ID: <20091029181249.GA26585@hecubus> Hi all, In hinting mode, where dom nodes like links are annotated with visible numbers (hints), and you are prompted to either type a number or substrings of link text, Conkeror is able to automatically select an element when the hints list has been narrowed down to a single possible match. How quickly it automatically selects this single match, thus exiting the hints mode interaction, is controlled by the variable hints_auto_exit_delay. The default is 500 milliseconds. If you set it to 0, that means it will never automatically select a match, and you must always hit return. Recently I was teaching someone about the hinting system, and how you can select a link by typing a substring of link text. The person said that they never used that feature because Conkeror would behave unpredictably if they either made a typo or typed too many characters. In the first instance, Conkeror might automatically follow the wrong link without the user realizing what they did wrong. In the second instance, Conkeror might automatically follow the right link, but the user could still typing because they were not being careful enough, and their extra keystrokes would run unintended commands. I think this person made a very good point. The default behavior should not put the user in the position of having to react to UI events that they did not intend to happen. We strive for Conkeror to have a "proactive" user interface, rather than a "reactive" one. Therefore I propose that the default value of hints_auto_exit_delay be changed to 0, so that by default, the user always has to hit return to end a hints interaction, and Conkeror's behavior with respect to this mode be 100% predictable and dependable. Shall I make this change? Thoughts? -- John Foerch From David.Kettler at dsto.defence.gov.au Thu Oct 29 15:32:01 2009 From: David.Kettler at dsto.defence.gov.au (David Kettler) Date: Fri, 30 Oct 2009 09:02:01 +1030 Subject: [Conkeror] hints_auto_exit_delay default [SEC=UNCLASSIFIED] In-Reply-To: <20091029181249.GA26585@hecubus> (John J. Foerch's message of "Thu, 29 Oct 2009 14:12:49 -0400") References: <20091029181249.GA26585@hecubus> Message-ID: > I think this person made a very good point. The default behavior should > not put the user in the position of having to react to UI events that they > did not intend to happen. We strive for Conkeror to have a "proactive" > user interface, rather than a "reactive" one. Therefore I propose that > the default value of hints_auto_exit_delay be changed to 0, so that by > default, the user always has to hit return to end a hints interaction, and > Conkeror's behavior with respect to this mode be 100% predictable and > dependable. > > Shall I make this change? Thoughts? I have hints_auto_exit_delay set to zero because I, too, find the auto behaviour to be unpredictable. There was peripheral comment on this in http://thread.gmane.org/gmane.comp.mozilla.conkeror/1184 regards, David. -- IMPORTANT: This email remains the property of the Australian Defence Organisation and is subject to the jurisdiction of section 70 of the CRIMES ACT 1914. If you have received this email in error, you are requested to contact the sender and delete the email. From jeremy at jeremyms.com Thu Oct 29 16:14:05 2009 From: jeremy at jeremyms.com (Jeremy Maitin-Shepard) Date: Thu, 29 Oct 2009 16:14:05 -0700 Subject: [Conkeror] hints_auto_exit_delay default In-Reply-To: <20091029181249.GA26585@hecubus> (John J. Foerch's message of "Thu, 29 Oct 2009 14:12:49 -0400") References: <20091029181249.GA26585@hecubus> Message-ID: <87ocnpkd36.fsf@jeremyms.com> "John J. Foerch" writes: > Shall I make this change? Thoughts? I must say, I've been using Conkeror with the default setting ever since this behavior was added, and I've never had a link be inadvertently followed due to this issue. Certainly, if there were no delay, it would be a problem. Note that as long as the user is still typing, the timeout will keep being reset and a link will not be followed, even if the matches had already been narrowed down to only one. This is done precisely to prevent unintended commands being executed. I think the issue is no so much what the default is but rather that people are unaware of exactly how the system works. In particular, you have to make sure not to pause too long if you make a mistake in typing before hitting backspace. I think it may require some adapting to learn how to use the system, and in particular it may be hard to adapt without actually understanding how it works, but overall following links automatically leads to a more efficient user interface, I think. For some people that prefer to type slower, simply setting the delay higher may work well (or disabling the auto following completely). I do generally agree that it is problematic when the effect of a key command is non-deterministic from the user's point of view, due to the possibility of some event occurring in between key strokes and causing the focus/context to change. The main case where this occurs in Conkeror is with the download prompt, but usually it is expected when it comes up, so it is not as much of an issue as it might otherwise be. I don't view the hints system as such a case, though; ruling out what the hints system does would essentially mean ruling out any "dynamic" (meaning time-dependent) behavior in user interfaces. I agree that there are some advantages to purely "static" user interfaces, but "dynamics" also can allow things to be more efficient, so in the end it is a tradeoff. -- Jeremy Maitin-Shepard From levy at msri.org Thu Oct 29 16:29:36 2009 From: levy at msri.org (Silvio Levy) Date: Thu, 29 Oct 2009 16:29:36 -0700 Subject: [Conkeror] hints_auto_exit_delay default In-Reply-To: Your message of Thu, 29 Oct 2009 16:14:05 -0700 Message-ID: <20091029232936.7F8EDDFCE@xi.msri.org> I like the current default behavior a lot. Jeremy's observations are right on target. Very occasionally I mistype and don't hit the backspace key in the 1-second window necessary to prevent a wrong link from being followed. Besides being rare, this event has always been inconsequential for me. By contrast, the time saved by not having to type return to follow each link is considerable. This said, so long as the current behavior is still supported somehow (even if it's not the default), I'm happy. Silvio From jjfoerch at earthlink.net Thu Oct 29 20:06:55 2009 From: jjfoerch at earthlink.net (John J. Foerch) Date: Thu, 29 Oct 2009 23:06:55 -0400 Subject: [Conkeror] hints_auto_exit_delay default In-Reply-To: <87ocnpkd36.fsf@jeremyms.com> References: <20091029181249.GA26585@hecubus> <87ocnpkd36.fsf@jeremyms.com> Message-ID: <20091030030655.GA9473@hecubus> On Thu, Oct 29, 2009 at 04:14:05PM -0700, Jeremy Maitin-Shepard wrote: > I think the issue is no so much what the default is but rather that > people are unaware of exactly how the system works. In particular, you > have to make sure not to pause too long if you make a mistake in typing > before hitting backspace. I think it may require some adapting to learn > how to use the system, and in particular it may be hard to adapt without > actually understanding how it works, but overall following links > automatically leads to a more efficient user interface, I think. For > some people that prefer to type slower, simply setting the delay higher > may work well (or disabling the auto following completely). > > I do generally agree that it is problematic when the effect of a key > command is non-deterministic from the user's point of view, due to the > possibility of some event occurring in between key strokes and causing > the focus/context to change. The main case where this occurs in > Conkeror is with the download prompt, but usually it is expected when it > comes up, so it is not as much of an issue as it might otherwise be. > > I don't view the hints system as such a case, though; ruling out what > the hints system does would essentially mean ruling out any "dynamic" > (meaning time-dependent) behavior in user interfaces. I agree that > there are some advantages to purely "static" user interfaces, but > "dynamics" also can allow things to be more efficient, so in the end it > is a tradeoff. Hi Jeremy, long time no see! I'm willing to make the case of the newbie here because it involves perhaps the most important and commonly used subsystem for all Conkeror users, regardless of their level of expertise, typing speed, browsing style, or what they want out of the program. We all need to be able to follow links with the keyboard. It does speak to the heart of the matter to say that people are unaware of exactly how the system works. Yet for such a basic and essential subsystem, should we expect them to? I would propose that changing the default to 0 would eliminate the need for the average user to learn exactly how the system works in order to use it at a basic level, and also to be confident in it as a dependable "black box". Apart from this feature, operating the hinting system is fairly obvious to anyone with any computer experience. We're all accustomed to hitting return at prompts in all variety of other softwares, so there will be no surprise there for anyone. For the person who wants no more out of Conkeror, they can just use it. Those of us who understand how it works don't mind putting a line in our config and forgetting about it, but those that don't will have their lives interrupted to "fix" it. Configuration should be about making the good better, not about making the [perceived] bad tolerable. Since hitting return at prompts is second nature to most of us, I don't think I am out on a limb to call it the base case in this situation. But just because something is the base case does not necessarily make it the best default. Speaking in generic terms, "Feature X" might be a better default, even though it has a steeper learning curve. To decide that matter, I would frame it with the following question: Can it be assumed that, despite a steeper learning curve, using Feature X is in all reasonable cases the preferable and obvious goal for the skilled user? Well, you know my answer to that one. :) Based not only on my own preference but also based on preferences of people I have talked to, such as the one that got me thinking about this again. Auto-selection is definitely more efficient for some people, but not everybody weighs the trade-off the same. Oh, I almost forgot a very important case. When a person is choosing a link via link text instead of via numbers, they may not have seen the link before starting the hints interaction, and they may need a chance to abort with C-g if an undesired link comes up instead of what they expected. "Typing blind" and seeing what comes up is also an efficiency gain, but doing so does not play nice with a timed-autoselect, and that is the main reason I turn off the feature in my own config. -- John Foerch From fredconcklin at gmail.com Thu Oct 29 20:45:45 2009 From: fredconcklin at gmail.com (fred concklin) Date: Thu, 29 Oct 2009 23:45:45 -0400 Subject: [Conkeror] hints_auto_exit_delay default In-Reply-To: <7bb8e7c30910292045j477e5482wdd7a8dc09e21be36@mail.gmail.com> References: <20091029181249.GA26585@hecubus> <7bb8e7c30910292045j477e5482wdd7a8dc09e21be36@mail.gmail.com> Message-ID: <7bb8e7c30910292045t7bf6ac1ckfc5d98b4f7a383e8@mail.gmail.com> I've never had a problem with the feature and find the default an attractive feature in conkeror. On Oct 29, 2009 2:13 PM, "John J. Foerch" wrote: Hi all, In hinting mode, where dom nodes like links are annotated with visible numbers (hints), and you are prompted to either type a number or substrings of link text, Conkeror is able to automatically select an element when the hints list has been narrowed down to a single possible match. How quickly it automatically selects this single match, thus exiting the hints mode interaction, is controlled by the variable hints_auto_exit_delay. The default is 500 milliseconds. If you set it to 0, that means it will never automatically select a match, and you must always hit return. Recently I was teaching someone about the hinting system, and how you can select a link by typing a substring of link text. The person said that they never used that feature because Conkeror would behave unpredictably if they either made a typo or typed too many characters. In the first instance, Conkeror might automatically follow the wrong link without the user realizing what they did wrong. In the second instance, Conkeror might automatically follow the right link, but the user could still typing because they were not being careful enough, and their extra keystrokes would run unintended commands. I think this person made a very good point. The default behavior should not put the user in the position of having to react to UI events that they did not intend to happen. We strive for Conkeror to have a "proactive" user interface, rather than a "reactive" one. Therefore I propose that the default value of hints_auto_exit_delay be changed to 0, so that by default, the user always has to hit return to end a hints interaction, and Conkeror's behavior with respect to this mode be 100% predictable and dependable. Shall I make this change? Thoughts? -- John Foerch _______________________________________________ Conkeror mailing list Conkeror at mozdev.org https://www.mozdev.org/mailman/listinfo/conkeror -------------- next part -------------- An HTML attachment was scrubbed... URL: