Re: [dev] surf rewrite for WebKit2GTK
On 10/25/14 13:41, F Hssn wrote: Following suckless's minimal philosophy, I'd be interested to find out if someone has done analysis on how an even minimal browser could be developed in terms of SLOC, since webkit (r172694) stands at ~2 million lines, 75% of which is C++, while webkitgtk-1.10.2 is ~1 million, 90% C++. I used cloc. I'm surprised webkitgtk is smaller than latest webkit. Do you really want to write your own Javascript engine?
[dev] [PATCH] [st] Fix issues with wcwidth() returning -1 for unsupported unicode chars
Hi suckless! First, thanks for st. Been using it for a long while,
still impressed at how it gets a lot of stuff right - stuff that urxvt
failed miserably at. There's only one issue that has been bothering me
particularly.
The issue itself:
Unicode characters added since unicode 5.2 (released in 2009, the
latest revision[1] is 7.0) are not supported by the wcwidth()
implementation of glibc, and as a result, they behave weirdly in st.
The man page of wcwidth() specifies that -1 is returned for invalid
unicode characters. I found a stack overflow question[2] about this
same issue.
How st handles it:
I made a gif[3] showing its behavior.
It just offsets the columns by the value returned by wcwidth,
expecting either 1 or 2, not -1. So each unsupported unicode character
behaves like a printable backspace.
Picked U+0524[4] for the tests. The st on the top shows the current
behavior, the st on the bottom is my patched version. The first two
lines typed in the gif are spaces followed by that character. Third
line is the letter 'a' just to show how it overlaps.
Then I used a tmux keybinding that is supposed to scan for URLs, but
the main effect here is refreshing the terminal contents, which makes
those characters vanish. That z^H is a typo, ignore that.
My patch:
Just wcwidth(...) -> abs(wcwidth(...))
In other words: if wcwidth returns -1, interpret that as a column
width of 1. It's a bit dirty and lazy, but it works wonderfully for
most characters.
I'm not sure what the "correct" solution would be, but it's definitely
not something as simple as this - would mean fixing the libc to
support unicode up to 7.0, or implementing our own version of it.
Opinions?
[1]: http://www.fileformat.info/info/unicode/version/index.htm
[2]:
http://stackoverflow.com/questions/16371418/why-does-wcwidth-return-1-with-a-sign-that-i-can-print-on-the-terminal
[3]: http://i.imgur.com/MDzMJJH.gif
[4]: http://www.fileformat.info/info/unicode/char/0524/index.htm
From 810937df2c8a693deb26ac278c73f0147353079b Mon Sep 17 00:00:00 2001
From: dequis
Date: Sat, 25 Oct 2014 23:07:38 -0300
Subject: [PATCH] Fix issues with wcwidth() returning -1 for unsupported
unicode chars
Unicode characters added since unicode 5.2 are not supported by the
wcwidth() implementation of glibc, and as a result, they behave weirdly
in st. The man page of wcwidth() specifies that -1 is returned for
invalid unicode characters.
This patch wraps the wcwidth() calls with abs() to ensure that a column
size of 1 is used for unknown unicode characters.
---
st.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/st.c b/st.c
index 23dd7f1..12af1ab 100644
--- a/st.c
+++ b/st.c
@@ -2576,7 +2576,7 @@ tputc(char *c, int len) {
unicodep = ascii = *c;
} else {
utf8decode(c, &unicodep, UTF_SIZ);
- width = wcwidth(unicodep);
+ width = abs(wcwidth(unicodep));
control = ISCONTROLC1(unicodep);
ascii = unicodep;
}
@@ -3440,7 +3440,7 @@ xdraws(char *s, Glyph base, int x, int y, int charlen, int bytelen) {
xp, winy + frc[i].font->ascent,
(FcChar8 *)u8c, u8cblen);
- xp += xw.cw * wcwidth(unicodep);
+ xp += xw.cw * abs(wcwidth(unicodep));
}
/*
--
2.1.2
Re: [dev] surf rewrite for WebKit2GTK
Seems like you deviated a bit from the thread subject.
Re: [dev] Macbook retina and dwm
> On Sun, 26 Oct 2014 01:00:05 +0300 Dimitris Zervas wrote:
> > I though about system wide font enlargement or passing other
> > resolution to X11 (possible?).
>
On Sun, Oct 26, 2014 at 12:41:15AM +0200, FRIGN wrote:
> You are close. The _real_ way to do it is set the DPI in xorg.conf
Or avoid touching that and put something like the following in
~/.x{initrc,profile} or the like:
xrandr --output DP-1 --dpi $DPI
pgpgzSgND9heO.pgp
Description: PGP signature
Re: [dev] Macbook retina and dwm
On Sun, 26 Oct 2014 01:00:05 +0300 Dimitris Zervas wrote: > Refer to this: https://wiki.archlinux.org/index.php/xorg#Display_size_and_DPI -- FRIGN
Re: [dev] Macbook retina and dwm
On Sun, 26 Oct 2014 01:00:05 +0300 Dimitris Zervas wrote: > Any ideas? > I though about system wide font enlargement or passing other resolution to > X11 (possible?). You are close. The _real_ way to do it is set the DPI in xorg.conf and enlargen the default font size. That's exactly what Apple does with HiDPI. Cheers FRIGN -- FRIGN
Re: [dev] surf rewrite for WebKit2GTK
2014-10-25 19:56 GMT-02:00 Wolfgang Corcoran-Mathe : >Any _popular_ > reboot would have to require people to change very little about > their habits, and to do no hard thinking about the nature of the > Web--if possible, no thinking at all. It doesn't need to be popular at first; it needs to be correct and simple and solve the problems at hand.
[dev] Macbook retina and dwm
Hello, I just bought my macbook pro retina 13". I am running Gentoo fairly well (a bit power hungry, I am open to suggestions, especially fan control). The main problem is the resolution. If I downscale with xrandr, it does bitmap scales so the fonts are horrible. Any ideas? I though about system wide font enlargement or passing other resolution to X11 (possible?).
Re: [dev] surf rewrite for WebKit2GTK
Quoth Daniel Camolês on Sat, Oct 25 2014 16:29 -0200: When the choice you have is between 500k or 2 million lines of code, it hurts to call anything suckless. I think the web needs a serious reboot. [snip] But when it comes to application distribution... I know that we have ssh and mainframe terminal applications, but those are not something most people are willing to use. Are you talking about what 'most people' want when you speak of a reboot? Because I doubt whether people who are afraid of _SSH_ (of all things) will choose a suite of less-sucking web applications if they can pervert Firefox into doing everything. Any _popular_ reboot would have to require people to change very little about their habits, and to do no hard thinking about the nature of the Web--if possible, no thinking at all. -- Wolfgang Corcoran-Mathe
Re: [dev] surf rewrite for WebKit2GTK
On Sat, Oct 25, 2014 at 06:12:29PM -0200, Daniel Camolês wrote: > The problem is not with this specific organization. I meant w3c as w3crap, as in all things www-related (sons of sgml, js, css, rdf, ...), not as the consortium, which imho isn't that much evil either: it has only the misfortune to aim to work on scripting languages and data representation in ways that imply you have to create disgusting nightmares of mad code to run it. Sorry if I put that wrong. -- Teodoro Santoni
Re: [dev] surf rewrite for WebKit2GTK
>> Am I alone? Is there any hope out there? > > TL;DR: Imho, there is none. W3c is the total opposite of suckless software. The problem is not with this specific organization. Design by committee is the source of the evil. Everybody wants to put a shiny new feature in there. Also, you never want to break backwards compatibility but you always need to do so a lot because the initial design was poor. So you add piles and piles of mess to compensate for that. > In either paths (interpret w3crap or deprecate it) I think interpreting w3crap will just generate piles of shit as you said. Breaking compatibility would be essential.
Re: [dev] surf rewrite for WebKit2GTK
On Sat, Oct 25, 2014 at 04:29:58PM -0200, Daniel Camolês wrote: > [snip] > I dream with the day when the Internet will be built > around a model simple and generic enough that a reasonable programmer > will be able to code a complete "browser" in a month of work or so. > > Am I alone? Is there any hope out there? TL;DR: Imho, there is none. W3c is the total opposite of suckless software. Let's start by the simple fact that this technology is based on sgml dialects and something you could call visual java. I can see only two strategies to make it suck less (in the suckless point of view, so the flexibility is maximum, it can fuck up your program and it's most likely user's fault or his/her responsibility to hack and patch it): kick NS plugins, cross-platform treatment of platform-specific things (like the codecs used for ) out of the code base and introducing better ways, nearer to what you see than what you get, to customize what you get from a given page (than JS userscripts and extensions, think about elinks scripting addons with an api that allows you to hack the DOM without workarounds like piping the page, or about w3m with a real GUI); and/or treat web pages as packages or apps, being able to compile them and even create overlays and repositories. Both the approaches are stupid, given the goals of w3 pages: the first is inconsistent and incredibly prone to Thompson hacks, the second is totally the opposite of what HTML is after the v4. What can we save? Either if it's by scripting or patching the source code, a suckless user shell for web pages (let's be honest: if we have to talk about web browser, thus developing something that renders more or less the same than firefox or chrome, one can only write a big pile of bloat just like gecko/blink/webkit, but in C) has to ship with a simpler and less standard-prone API for manipulating what happens on the screen when parsing a DOM, and maybe should have a static compiler for javascript, to C or to the scripting language for the "browser". That would break, however, the classic workflow a user has with a browser. This is a rant based on "mantaining the status quo but having something that doesn't coredump on google maps", one can instead do something lesser dumb and adding the modern capabilities to dumb terminal logic. Nothing impedes to create extensions to VT1XX, for images, GL, videos, and replace cgi things with, for example, 9p-style interaction. Or replacing the actual html, css and js files in webservers with tcl scripts or things like that. Something you would download, with its corresponding media files, and feed to a wish shell or similar software. In either paths (interpret w3crap or deprecate it) lots of work may be involved. And may end with another shitty bloated system. We are talking about taking over the world wide web, after all, lol. -- Teodoro Santoni
Re: [dev] surf rewrite for WebKit2GTK
> No, that was not what I was talking about. In this browser discussion, > I am not interested in ways to understand complexity as your proposed > solution. I think the fact that the browser needs to be this huge is a > signal that things are very, very wrong, and a new solution for > application distribution needs to come. I'm still struggling to communicate this idea (everywhere, not just here). The browser is complex because it relies on eighteen layers of dependencies rather than just one. And the layers of dependencies are caused by wanting some fuzzy sense of backwards compatibility rather than digging into exactly what services one requires. That is my candidate answer to your question of what is "very, very wrong" and what the new solution to distribution needs to be. I hope that white-box tests will allow more fine-grained code sharing without worrying about compatibility. That will eliminate compounding drag across many layers of dependencies. This may be more radical than you expect. But the truth is often not reassuring. "A prematurely frozen library doesn't just force people to live with it. People react to it by wrapping it in a cleaner interface. But then they prematurely freeze the new interface, and it starts accumulating warts and bolt-on features just like the old one. Now you have two interfaces. Was forking the existing interface really so much worse an alternative? How much smaller might each codebase in the world be without all the combinatorial explosion of interfaces wrapping other interfaces" -- http://akkartik.name/post/libraries2
Re: [dev] surf rewrite for WebKit2GTK
2014-10-25 16:59 GMT-02:00 Kartik Agaram : >> Am I alone? Is there any hope out there? > > I'm working on this problem. I think the solution is white-box testing > from the ground up, so that software encodes not just the rules about > what to do but the specific scenarios that the programmer considered. No, that was not what I was talking about. In this browser discussion, I am not interested in ways to understand complexity as your proposed solution. I think the fact that the browser needs to be this huge is a signal that things are very, very wrong, and a new solution for application distribution needs to come. I am not trying to understand the current browser's code nor looking for ways to deal with huge messes. I can't believe the the web and the browser are the better solution for this problem.
Re: [dev] surf rewrite for WebKit2GTK
On Sat, Oct 25, 2014 at 04:29:58PM -0200, Daniel Camolês wrote: > When the choice you have is between 500k or 2 million lines of code, > it hurts to call anything suckless. I think the web needs a serious > reboot. It started out as a markup language for presentation-only, and > then it was morphed through a convoluted series of additions into an > application distribution plataform. The browser is the dumb terminal > of our days. Today the web browser is the most awful piece of software > I need to use. I dream with the day when the Internet will be built > around a model simple and generic enough that a reasonable programmer > will be able to code a complete "browser" in a month of work or so. > > Am I alone? Is there any hope out there? I remember reading something Russ Cox said about plan9port being created because it would be easier to port all the Plan 9 userland to Unix than to port a full-featured web browser to Plan 9.
Re: [dev] surf rewrite for WebKit2GTK
> Am I alone? Is there any hope out there? I'm working on this problem. I think the solution is white-box testing from the ground up, so that software encodes not just the rules about what to do but the specific scenarios that the programmer considered. Because we only encode rules and not intentions we paint ourselves into corners where it's always easier to add more layers of code than it is to rework existing layers. But this is hard to do with the existing stack, because unix was invented before we appreciated the power of tests, and it makes it too hard to write tests for blocking IO, or for the 'find' command, or for what happens when a program runs out of memory or disk. What we need is a simple OS and a comprehensive library of fakes (https://en.wikipedia.org/wiki/Mock_object#Mocks.2C_fakes_and_stubs) for all OS concepts. This is what I'm working on. http://akkartik.name/about http://akkartik.name/post/readable-bad https://news.ycombinator.com/item?id=8308881 https://news.ycombinator.com/item?id=8309344 https://news.ycombinator.com/item?id=8327008 https://news.ycombinator.com/item?id=8375219 http://www.ribbonfarm.com/2014/04/09/the-legibility-tradeoff
Re: [dev] surf rewrite for WebKit2GTK
When the choice you have is between 500k or 2 million lines of code, it hurts to call anything suckless. I think the web needs a serious reboot. It started out as a markup language for presentation-only, and then it was morphed through a convoluted series of additions into an application distribution plataform. The browser is the dumb terminal of our days. Today the web browser is the most awful piece of software I need to use. I dream with the day when the Internet will be built around a model simple and generic enough that a reasonable programmer will be able to code a complete "browser" in a month of work or so. Am I alone? Is there any hope out there? Document distribution seems easy. Hyperlinking was an wonderful idea, but it's not complicated. But when it comes to application distribution. By application distribution I mean, when we want to develop and maintain software in a central location and enable several users with a generic client to use it. I know that we have ssh and mainframe terminal applications, but those are not something most people are willing to use. Answering your question about netsurf, I think the point is that it doesn't currently run javascript. I think that is good thing, but most site developers in the world doesn't. Also, netsurf developers also intend to implement javascript at some point. But today for a lot of sites you have to stick with one of the three main rendering engines. 2014-10-25 15:41 GMT-02:00 F Hssn : > On Thu, Oct 23, 2014 at 6:54 AM, Quentin Rameau wrote: >> Hi, I tried to port surf for the webkit2 (WebKitGTK 2.6 / GTK3), here >> is the code: >> git://quinq.eu.org/surf2 >> There will be some bugs, feel free to try it, feedback welcomed. > > Noob question (joined the list a couple months ago), > > Since surf, uzbl, jumanji etc are minimal browsers in terms of UI's, > is the GTK version of webkit (as opposed to directly using Webkit) > used primarily for providing widget functionality to HTML? (so that > widgets can be embedded in a webpage). > > Following suckless's minimal philosophy, I'd be interested to find out > if someone has done analysis on how an even minimal browser could be > developed in terms of SLOC, since webkit (r172694) stands at ~2 > million lines, 75% of which is C++, while webkitgtk-1.10.2 is ~1 > million, 90% C++. I used cloc. I'm surprised webkitgtk is smaller than > latest webkit. > > Furthermore, since webkit is C++, I found (through wikipedia) a > project called netsurf, which is a browser written in C (~0.5 million > total, 50% is C). I wonder if anyone has considered that. > > Thanks and regards, > FH >
Re: [dev] surf rewrite for WebKit2GTK
On Thu, Oct 23, 2014 at 6:54 AM, Quentin Rameau wrote: > Hi, I tried to port surf for the webkit2 (WebKitGTK 2.6 / GTK3), here > is the code: > git://quinq.eu.org/surf2 > There will be some bugs, feel free to try it, feedback welcomed. Noob question (joined the list a couple months ago), Since surf, uzbl, jumanji etc are minimal browsers in terms of UI's, is the GTK version of webkit (as opposed to directly using Webkit) used primarily for providing widget functionality to HTML? (so that widgets can be embedded in a webpage). Following suckless's minimal philosophy, I'd be interested to find out if someone has done analysis on how an even minimal browser could be developed in terms of SLOC, since webkit (r172694) stands at ~2 million lines, 75% of which is C++, while webkitgtk-1.10.2 is ~1 million, 90% C++. I used cloc. I'm surprised webkitgtk is smaller than latest webkit. Furthermore, since webkit is C++, I found (through wikipedia) a project called netsurf, which is a browser written in C (~0.5 million total, 50% is C). I wonder if anyone has considered that. Thanks and regards, FH
Re: [dev] [surf] user certificates
Hi, in WebKit2 API, there is webkit_web_context_allow_tls_certificate_for_host() [1]. Maybe that can help you. But I think that certificates aren't directly handled by WebKit and you should manage them with other tools for your users. [1]: http://webkitgtk.org/reference/webkit2gtk/2.4.2/WebKitWebContext.html#webkit-web-context-allow-tls-certificate-for-host On Mon, May 5, 2014 at 10:18 PM, Markus Teich wrote: > Heyho, > > how to implement user certificates? I did not find any hint in the > documentation > of webkit-gtk or libsoup… :( > > --Markus >
