Re: [9fans] Links for Plan 9

Re: [9fans] Links for Plan 9

[YAMANASHI Takeshi]

Links not only suicide itself, also causes kernel panic if
`-http-proxy' option is specified.

	term% links -http-proxy xx.xx.xx.xx:8080
	links 138: suicide: sys: trap: fault read addr=0x0 pc=0x000dafd6

	panic: boot process died: exit
	ktrace /kernel/path 80106720 800238bc
	 :
	 :

You can see more complete messagase here;
	http://p9c.cc.titech.ac.jp/plan9/tmp/vmscr.bmp
--
YAMANASHI Takeshi

Re: [9fans] Links for Plan 9

[YAMANASHI Takeshi]

On Tue Dec 16 10:28:42 JST 2003, okamoto@granite.cias.osakafu-u.ac.jp wrote:
> By the way, how about to have a "exit/quit" command in
> fossilcons(8) to make return the console i/o to auth
> server.

You can quit from `con -l /srv/fscons' command by typing
Control-\ and then `q' for quit.  See con(1) for more details.
--
YAMANASHI Takeshi

Re: [9fans] Links for Plan 9

[okamoto]

>> fossilcons(8) to make return the console i/o to auth
>> server.
>
> You can quit from `con -l /srv/fscons' command by typing
> Control-\ and then `q' for quit.  See con(1) for more details.

Ok, I'll try that.

By the way, is there drawterm for 9p2000?

Kenji

Re: [9fans] Links for Plan 9

[YAMANASHI Takeshi]

> 	http://pages.cpsc.ucalgary.ca/~mirtchov/p9/links/

Links seems to handle UTF-8 pages correctly. try;
	hget http://p9c.cc.titech.ac.jp/plan9/tmp/links.gz | gunzip | page -w

A good news for japanese-reading people  :).

However, the fonts used by links now is not
very good and the star character appears
in the place of kanji character.
--

Re: [9fans] Links for Plan 9

[mirtchov]

>> 	http://pages.cpsc.ucalgary.ca/~mirtchov/p9/links/
>
> Links seems to handle UTF-8 pages correctly. try;
> 	hget http://p9c.cc.titech.ac.jp/plan9/tmp/links.gz | gunzip | page -w
>
> A good news for japanese-reading people  :).
>
> However, the fonts used by links now is not
> very good and the star character appears
> in the place of kanji character.
> --

also note: there is a binary image on the web now, for those of you
with a 386 architecture but without the latest ape/libdraw or ability
to pull one.

there are also numerous fixes to mkfiles (as reported by Presotto) and
leftover binaries caused by my carelessness.

the screenshots that you see on my web page (the slashdot and google
one) show a bug in drawing lines and filled rectangles -- the ordering
of bytes is BGR instead of RGB, this could be just my choice of
incorrect get_color routine.  so everything that is not a bitmap (such
as text and images) was drawn with swapped red and blue values.

this bug is now fixed, so you can look at slashdot in all its
(somewhat more properly rendered) glory :)

andrey

ps: the screenshot on titech.ac.jp was taken with an old binary, so
you should probably download it again.

[9fans] Links for Plan 9

[andrey mirtchovski]

After Russ' much needed creation of ape/libdraw I managed to get the Links
graphical web browser ported to Plan 9 and working in a somewhat useable
state.

Download the code, see screenshots, bugs, what's there and what's missing
information and others at the web site:

	http://pages.cpsc.ucalgary.ca/~mirtchov/p9/links/

There are several things unfinished, most notably JavaScript support (turned
off due to problems with sites like google) and a bug with select()
descriptors which causes sites like cnn.com, nytimes (every time) and
theregister (on occasion) to fail loading.

The links code is horrendous, with most of the comments written in Czech
(funny, in the few places where I understand the comments they seem to be
horrendous too) with the name Jim Beam interspersed just so foreign readers
would know what's hapening.

The plan9.c (graphics driver for Plan 9) code, for which I am responsible,
isn't much better. Luckily acme takes the edge off and after an hour or two
looking at the code you'll be able to go bug-hunting. I'm letting it go
precisely with that in mind -- to get a broader audience looking for the
bugs. I won't have that much time to do it for a while...

andrey

Re: [9fans] Links for Plan 9

[David Presotto]

You have a .8 and a .a8 in png/mkfile

Re: [9fans] Links for Plan 9

[David Presotto]

Also, there's a bug in the ape mips ape/lib/ap/lock.c.  New one on sources
though the scan won't get it for a while.

Re: [9fans] Links for Plan 9

[John Stalker]

> The links code is horrendous, with most of the comments written in Czech
> (funny, in the few places where I understand the comments they seem to be
> horrendous too) with the name Jim Beam interspersed just so foreign readers
> would know what's hapening.
>
> andrey

Thanks for porting it.  Despite its various peculiarities it is the
main browser I use on linux and FreeBSD.

Actually many of the comments are in English.   The ones in Czech are
indeed in the kind of Czech that would make a grammar teacher
blush.  Or inflict bodily injury on the student.  Depends on the
teacher.  I set about translating them, toning down some of the more
extreme bits.  At my current rate it would take about 20-30 hours to
do the whole source.  I will almost certainly lose interest before
then, so I should be more selective.  I did javascript.c--not a good
test case since most of the source code is the number 2--and part of
ipret.c.  I assume that javascr.c and jsint.c are also of interest
since the javascript part of the port is not working.  What other files
are you interested in?

The debugging messages are also in a mix of languages as are variable
names.  I have been ignoring both.  Translating variable names is
not a good idea, but if you are curious about any debugging messages
just email me.

-
John Stalker
Department of Mathematics
Princeton University
(609)258-6469

Re: [9fans] Links for Plan 9

[mirtchov]

> Thanks for porting it.  Despite its various peculiarities it is the
> main browser I use on linux and FreeBSD.
>
> Actually many of the comments are in English.   The ones in Czech are
> indeed in the kind of Czech that would make a grammar teacher
> blush.  Or inflict bodily injury on the student.  Depends on the
> teacher.  I set about translating them, toning down some of the more
> extreme bits.  At my current rate it would take about 20-30 hours to
> do the whole source.  I will almost certainly lose interest before
> then, so I should be more selective.  I did javascript.c--not a good
> test case since most of the source code is the number 2--and part of
> ipret.c.  I assume that javascr.c and jsint.c are also of interest
> since the javascript part of the port is not working.  What other files
> are you interested in?
>
> The debugging messages are also in a mix of languages as are variable
> names.  I have been ignoring both.  Translating variable names is
> not a good idea, but if you are curious about any debugging messages
> just email me.

Bell-Labs should decide whether they want the browser on sources.  If
not, I will set up a replica here so that I can accept diffs more
easily and won't require people to download 3.3MB .tgz file for each
little change to the source.

Many 9fans have expressed desire to help make the browser more Plan
9-friendly (things like plumbing and copy/paste for example), so the
usefulness of Links in Plan 9 (at least currently) seems to outweigh
its design mishaps, so there is a high possibility that the source
will be looked at frequently.  Making the comments readable, then, is
a very good idea, thanx.

As for having to modify every line in every file -- I feel your
pain...  A major part of doing the port involved having to add proper
casts between unsigned and signed characters pretty much everywhere.
The guys who wrote it insist that characters be unsigned, APE insists
that strlen() be passed a signed char.

In just one file (jsint.c) I count 200+ occasions of:

3731c3731
< 	if (!js_ctx)internal("js_upcall_get_subframes called with NULL context pointer\n");
---
> 	if (!js_ctx)internal((unsigned char *)"js_upcall_get_subframes called with NULL context pointer\n");

andrey

Re: [9fans] Links for Plan 9

[David Presotto]

[-- Attachment #1: Type: text/plain, Size: 59 bytes --]

I'm happier with you doing the distribution of the browser.

[-- Attachment #2: Type: message/rfc822, Size: 4033 bytes --]

From: mirtchov@cpsc.ucalgary.ca
To: 9fans@cse.psu.edu
Subject: Re: [9fans] Links for Plan 9
Date: Fri, 12 Dec 2003 10:13:16 -0700
Message-ID: <b9f36269cf426cdd70fc8f819fd0b6ad@plan9.ucalgary.ca>

> Thanks for porting it.  Despite its various peculiarities it is the
> main browser I use on linux and FreeBSD.
>
> Actually many of the comments are in English.   The ones in Czech are
> indeed in the kind of Czech that would make a grammar teacher
> blush.  Or inflict bodily injury on the student.  Depends on the
> teacher.  I set about translating them, toning down some of the more
> extreme bits.  At my current rate it would take about 20-30 hours to
> do the whole source.  I will almost certainly lose interest before
> then, so I should be more selective.  I did javascript.c--not a good
> test case since most of the source code is the number 2--and part of
> ipret.c.  I assume that javascr.c and jsint.c are also of interest
> since the javascript part of the port is not working.  What other files
> are you interested in?
>
> The debugging messages are also in a mix of languages as are variable
> names.  I have been ignoring both.  Translating variable names is
> not a good idea, but if you are curious about any debugging messages
> just email me.

Bell-Labs should decide whether they want the browser on sources.  If
not, I will set up a replica here so that I can accept diffs more
easily and won't require people to download 3.3MB .tgz file for each
little change to the source.

Many 9fans have expressed desire to help make the browser more Plan
9-friendly (things like plumbing and copy/paste for example), so the
usefulness of Links in Plan 9 (at least currently) seems to outweigh
its design mishaps, so there is a high possibility that the source
will be looked at frequently.  Making the comments readable, then, is
a very good idea, thanx.

As for having to modify every line in every file -- I feel your
pain...  A major part of doing the port involved having to add proper
casts between unsigned and signed characters pretty much everywhere.
The guys who wrote it insist that characters be unsigned, APE insists
that strlen() be passed a signed char.

In just one file (jsint.c) I count 200+ occasions of:

3731c3731
< 	if (!js_ctx)internal("js_upcall_get_subframes called with NULL context pointer\n");
---
> 	if (!js_ctx)internal((unsigned char *)"js_upcall_get_subframes called with NULL context pointer\n");

andrey

Re: [9fans] Links for Plan 9

[Dan Cross]

David Presotto <presotto@closedmind.org> writes:
> I'm happier with you doing the distribution of the browser.

It seems like something generally useful enough that it'd make sense to
put on sources under /contrib....  I urge you to reconsider.  One sources
for things should be enough.

	- Dan C.

Re: [9fans] Links for Plan 9

[Russ Cox]

> David Presotto <presotto@closedmind.org> writes:
> > I'm happier with you doing the distribution of the browser.
>
> It seems like something generally useful enough that it'd make sense to
> put on sources under /contrib....  I urge you to reconsider.  One sources
> for things should be enough.

I agree with Dave.  The browser is disgusting.  It looks ugly
on the outside, and it's even uglier on the inside, in many ways.
About the only thing it has going for it is it can display web
pages better than mothra.

If the people here clean it up and make it look and behave
like a Plan 9 program, then great.  But until then, no.

Russ

P.S. Next you're going to complain that there are things
in /extra that don't look and behave much like Plan 9 programs,
but Links is much further removed from those.

Re: [9fans] Links for Plan 9

[Dan Cross]

"Russ Cox" <rsc@swtch.com> writes:
> P.S. Next you're going to complain that there are things
> in /extra that don't look and behave much like Plan 9 programs,
> but Links is much further removed from those.

Close, but no.  Now I'm going to complain that a web browser is the
*first thing* many, many new users ask for.  Links is, arguably, the
first *viable* browser that's compiled and run on Plan 9.  It may not
be pretty, inside or out (and the latter is certainly your opinion),
but at least it works, which is more than can be said for, e.g., Mothra
or i.

Do you really think it's wise to ignore that?

	- Dan C.

Re: [9fans] Links for Plan 9

[David Presotto]

ignorance is my strong suit...

I'ld rather the only person that understands the thing (however
vageuly) be the one to maintain and distribute it, at least till
it cleans up a lot more.

Re: [9fans] Links for Plan 9

[Skip Tavakkolian]

> Links is, arguably, the
> first *viable* browser that's compiled and run on Plan 9.

It would be a very tenuous argument.

BTW, why doesn't Charon count?

Re: [9fans] Links for Plan 9

[ron minnich]

Why not take this as an opportunity to build the Plan 9 equivalent of the
FreeBSD "ports" collection. You could do a much better job with Plan 9
anyway.

ron

Re: [9fans] Links for Plan 9

[mirtchov]

> Why not take this as an opportunity to build the Plan 9 equivalent of the
> FreeBSD "ports" collection. You could do a much better job with Plan 9
> anyway.
>
> ron

I'll make a replica for ports of lunix software at ucalgary next week.
I have an extra disk I can stick in for that purpose.  The main issue
then is whether to use sources' auth information for the ports too...

Re: [9fans] Links for Plan 9

[ron minnich]

On Sat, 13 Dec 2003 mirtchov@cpsc.ucalgary.ca wrote:

> I'll make a replica for ports of lunix software at ucalgary next week.
> I have an extra disk I can stick in for that purpose.  The main issue
> then is whether to use sources' auth information for the ports too...


this is worth working out. One of the truly wonderful things about FreeBSD
is the ability to:
cd /usr/ports/whatever
make


and have it all work right. Allows one to not think at all, and who can
complain about that?

ron

Re: [9fans] Links for Plan 9

[boyd, rounin]

> and have it all work right. Allows one to not think at all, and who can
> complain about that?

have you been reading the freebsd mailing list?  ouch!

Re: [9fans] Links for Plan 9

[ron minnich]

On Sun, 14 Dec 2003, boyd, rounin wrote:

> > and have it all work right. Allows one to not think at all, and who can
> > complain about that?
>
> have you been reading the freebsd mailing list?  ouch!


No, I left it several years ago. But, /usr/ports is a great thing.

ron

Re: [9fans] Links for Plan 9

[boyd, rounin]

i think i woulda taken mothra and smashed whatever was needed into it.

duff knew his graphics.

Re: [9fans] Links for Plan 9

[ron minnich]

On Sun, 14 Dec 2003, boyd, rounin wrote:

> i think i woulda taken mothra and smashed whatever was needed into it.

I still don't know the source rules around mothra.

ron

Re: [9fans] Links for Plan 9

[Dan Cross]

mirtchov@cpsc.ucalgary.ca writes:
> I'll make a replica for ports of lunix software at ucalgary next week.

And thus begins the slow painful process of the Balkanization of Plan 9.
Started by a Bulgarian, no less!

> I have an extra disk I can stick in for that purpose.  The main issue
> then is whether to use sources' auth information for the ports too...

Yes, do, if Bell Labs will let you.  My biggest concern is really the
auth stuff, not where the data actually comes from.

	- Dan C.

Re: [9fans] Links for Plan 9

[Dan Cross]

"Skip Tavakkolian" <fst@centurytel.net> writes:
> BTW, why doesn't Charon count?

I don't include Charon because it's *really* an Inferno program, and
what's more, every time I try to use it, I just become incredibly
frustrated and breath a sigh of relief when I can use something else.

	- Dan C.

Re: [9fans] Links for Plan 9

[okamoto]

> I agree with Dave.  The browser is disgusting.  It looks ugly
> on the outside, and it's even uglier on the inside, in many ways.
> About the only thing it has going for it is it can display web
> pages better than mothra.

I feel it's cute on the outside.  This is depend on what the feeling
one has.   I feel it cute, because it has no decorative excess menu-like
things at first, very simple.

On the inside, I have not take a look, so I request you to be more
specific on this point.   If not, you may be counted as accusing them.

> P.S. Next you're going to complain that there are things
> in /extra that don't look and behave much like Plan 9 programs,
> but Links is much further removed from those.

ditto.

Kenji

Re: [9fans] Links for Plan 9

[Gorka Guardiola Múzquiz]

> On the inside, I have not take a look, so I request you to be more
> specific on this point.   If not, you may be counted as accusing them.

I am hacking it to try to add plumber and I feel the same as him.

An example:


	if (!a) {
		if (!anonymous) msg_box(ses->term, getml(m1, NULL), TEXT(T_UNKNOWN_TYPE), AL_CENTER | AL_EXTD_TEXT, TEXT(T_CONTEN_TYPE_IS), " ", m1, ".\n", TEXT(T_DO_YOU_WANT_TO_SAVE_OR_DISLPAY_THIS_FILE), NULL, ses, 3, TEXT(T_SAVE), tp_save, B_ENTER, TEXT(T_DISPLAY), tp_display, 0, TEXT(T_CANCEL), tp_cancel, B_ESC);
		else msg_box(ses->term, getml(m1, NULL), TEXT(T_UNKNOWN_TYPE), AL_CENTER | AL_EXTD_TEXT, TEXT(T_CONTEN_TYPE_IS), " ", m1, ".\n", TEXT(T_DO_YOU_WANT_TO_SAVE_OR_DISLPAY_THIS_FILE), NULL, ses, 2, TEXT(T_DISPLAY), tp_display, B_ENTER, TEXT(T_CANCEL), tp_cancel, B_ESC);
	} else {
		m2 = stracpy(a[0].label ? a[0].label : (unsigned char *)"");
		if (!anonymous) msg_box(ses->term, getml(m1, m2, NULL), TEXT(T_WHAT_TO_DO), AL_CENTER | AL_EXTD_TEXT, TEXT(T_CONTEN_TYPE_IS), " ", m1, ".\n", TEXT(T_DO_YOU_WANT_TO_OPEN_FILE_WITH), " ", m2, ", ", TEXT(T_SAVE_IT_OR_DISPLAY_IT), NULL, ses, 4, TEXT(T_OPEN), tp_open, B_ENTER, TEXT(T_SAVE), tp_save, 0, TEXT(T_DISPLAY), tp_display, 0, TEXT(T_CANCEL), tp_cancel, B_ESC);
		else msg_box(ses->term, getml(m1, m2, NULL), TEXT(T_WHAT_TO_DO), AL_CENTER | AL_EXTD_TEXT, TEXT(T_CONTEN_TYPE_IS), " ", m1, ".\n", TEXT(T_DO_YOU_WANT_TO_OPEN_FILE_WITH), " ", m2, ", ", TEXT(T_SAVE_IT_OR_DISPLAY_IT), NULL, ses, 3, TEXT(T_OPEN), tp_open, B_ENTER, TEXT(T_DISPLAY), tp_display, 0, TEXT(T_CANCEL), tp_cancel, B_ESC);
	}


				G

Re: [9fans] Links for Plan 9

[okamoto]

> An example:
> 
> 
> 	if (!a) {
> 		if (!anonymous) msg_box(ses->term, getml(m1, NULL), TEXT(T_UNKNOWN_TYPE), AL_CENTER | AL_EXTD_TEXT, TEXT(T_CONTEN_TYPE_IS), " ", m1, ".\n", TEXT(T_DO_YOU_WANT_TO_SAVE_OR_DISLPAY_THIS_FILE), NULL, ses, 3, TEXT(T_SAVE), tp_save, B_ENTER, TEXT(T_DISPLAY), tp_display, 0, TEXT(T_CANCEL), tp_cancel, B_ESC);
> 		else msg_box(ses->term, getml(m1, NULL), TEXT(T_UNKNOWN_TYPE), AL_CENTER | AL_EXTD_TEXT, TEXT(T_CONTEN_TYPE_IS), " ", m1, ".\n", TEXT(T_DO_YOU_WANT_TO_SAVE_OR_DISLPAY_THIS_FILE), NULL, ses, 2, TEXT(T_DISPLAY), tp_display, B_ENTER, TEXT(T_CANCEL), tp_cancel, B_ESC);
> 	} else {
> 		m2 = stracpy(a[0].label ? a[0].label : (unsigned char *)"");
> 		if (!anonymous) msg_box(ses->term, getml(m1, m2, NULL), TEXT(T_WHAT_TO_DO), AL_CENTER | AL_EXTD_TEXT, TEXT(T_CONTEN_TYPE_IS), " ", m1, ".\n", TEXT(T_DO_YOU_WANT_TO_OPEN_FILE_WITH), " ", m2, ", ", TEXT(T_SAVE_IT_OR_DISPLAY_IT), NULL, ses, 4, TEXT(T_OPEN), tp_open, B_ENTER, TEXT(T_SAVE), tp_save, 0, TEXT(T_DISPLAY), tp_display, 0, TEXT(T_CANCEL), tp_cancel, B_ESC);
> 		else msg_box(ses->term, getml(m1, m2, NULL), TEXT(T_WHAT_TO_DO), AL_CENTER | AL_EXTD_TEXT, TEXT(T_CONTEN_TYPE_IS), " ", m1, ".\n", TEXT(T_DO_YOU_WANT_TO_OPEN_FILE_WITH), " ", m2, ", ", TEXT(T_SAVE_IT_OR_DISPLAY_IT), NULL, ses, 3, TEXT(T_OPEN), tp_open, B_ENTER, TEXT(T_DISPLAY), tp_display, 0, TEXT(T_CANCEL), tp_cancel, B_ESC);
> 	}

I insist that THIS IS NOT ESSENTIAL, JUST DIFFICULT TO READ THING.☺

Kenji

Re: [9fans] Links for Plan 9

[Skip Tavakkolian]

> I insist that THIS IS NOT ESSENTIAL

As boyd would probably attest, this argument could also be made about
gun locks.

Re: [9fans] Links for Plan 9

[boyd, rounin]

> As boyd would probably attest, this argument could also be made about
> gun locks.

perhaps not essential [trigger locks], but it is the LAW here.

the M9 safety/de-cock is very nice.

here is the last Tango who 'argued' with me:

   http://www.insultant.net/targets/last.jpg

Re: [9fans] Links for Plan 9

[Geoff Collyer]

The inside of links is like bad undergraduate code: awful style,
duplicated code, everything done about as poorly as it could be,
overly complicated, superstitious code (call this routine multiple
times, just in case of some random, undescribed event), massive
function argument lists, inappropriate use of data structures.  I've
worked it over this weekend so that it doesn't consume all your CPU
time, but it's still mysteriously slow on some pages (e.g.,
nytimes.com).

Re: [9fans] Links for Plan 9

[okamoto]

> but it's still mysteriously slow on some pages (e.g.,
> nytimes.com).

yeah, and we have to resize the widow to see it...
Have you found the solution of it?

Kenji

Re: [9fans] Links for Plan 9

[Geoff Collyer]

I haven't had to resize my links window; can you describe your problem
in more detail?

I think I've figured out how to cure the slowness, though.

Re: [9fans] Links for Plan 9

[okamoto]

> I haven't had to resize my links window; can you describe your problem
> in more detail?

To see www.nytimes.com, it takes long time to get that page.
However, we can get it's part fasters when we resize the links window
and hit Ins key.
Without this, it took over ten minutes!

Kenji

Re: [9fans] Links for Plan 9

[Geoff Collyer]

nytimes.com no longer takes 10 minutes from here (California).  There
were also problems with flushing graphics.

Re: [9fans] Links for Plan 9

[okamoto]

> nytimes.com no longer takes 10 minutes from here (California).  There
> were also problems with flushing graphics.

Great!
Working for links now is very nice to all of us.
I'm now in distance from it, because I have trouble to set-up
fossil+venti system at home... sigh

Kenji

Re: [9fans] Links for Plan 9

[okamoto]

> I'm now in distance from it, because I have trouble to set-up
> fossil+venti system at home... sigh

I got the solution, that is, DON'T EDIT /adm/users by hand!
Now, I have fossil+venti+auth file server + terminals network
also in my home.

By the way, how about to have a "exit/quit" command in
fossilcons(8) to make return the console i/o to auth
server.   I know most people are using rio for Venti+Fossil+Auth
server, however, it's also possible not to use rio on that machine
just to save memory.   Is this a ridiculous idea?

Kenji

Re: [9fans] Links for Plan 9

[Russ Cox]

> > I agree with Dave.  The browser is disgusting.  It looks ugly
> > on the outside, and it's even uglier on the inside, in many ways.
> > About the only thing it has going for it is it can display web
> > pages better than mothra.
>
> I feel it's cute on the outside.  This is depend on what the feeling
> one has.   I feel it cute, because it has no decorative excess menu-like
> things at first, very simple.

As far as outside appearances go, I wish it used fonts
that fit in more with the rest of Plan 9.  That's my main gripe.
The X11 fixed fonts (which it seems to have hard-coded in
various sizes) have always been ugly.

I do think that Links is a demonstration that we could manage
to get a small web browser up and running without too much
code.  I'm not too optimistic though, since I fear style sheets
and the like will soon obsolete any such attempt just like tables
obsoleted mothra.

> On the inside, I have not take a look, so I request you to be more
> specific on this point.   If not, you may be counted as accusing them.

If you take a look, you'll see that I don't need to be any more
specific.  Everything about the code is inscrutable.

Russ

Re: [9fans] Links for Plan 9

[okamoto]

> As far as outside appearances go, I wish it used fonts
> that fit in more with the rest of Plan 9.  That's my main gripe.
> The X11 fixed fonts (which it seems to have hard-coded in
> various sizes) have always been ugly.

You mean menu and staus line fonts not text fonts?
For text fonts in the main page I don't see any problem.

> I do think that Links is a demonstration that we could manage
> to get a small web browser up and running without too much
> code.

This is the very point which hit my brain!

>I'm not too optimistic though, since I fear style sheets
> and the like will soon obsolete any such attempt just like tables
> obsoleted mothra.

However, we should have one not complaining its difficulties.

> If you take a look, you'll see that I don't need to be any more
> specific.  Everything about the code is inscrutable.

I believe you.   However, this is also shown to the public not
only to mailing-list.

Kenji

Re: [9fans] Links for Plan 9

[mirtchov]

> As far as outside appearances go, I wish it used fonts
> that fit in more with the rest of Plan 9.  That's my main gripe.

I've listed this on the download page as one of the first things to
implement if one decides to dig in.

Gorka has found something a bit more important -- plumbing of unknown
link types -- which I will put on the web tomorrow.

> I'm not too optimistic though, since I fear style sheets
> and the like will soon obsolete any such attempt just like tables
> obsoleted mothra.

We may be in luck here actually -- we have the infinite monkeys
working for us.  All we need to do is keep spending (conceivably less)
time porting their efforts.

andrey

Re: [9fans] Links for Plan 9

[John Stalker]

sic scripsit Russ Cox die XIV Decembris anni MMIII:

> I do think that Links is a demonstration that we could manage
> to get a small web browser up and running without too much
> code.  I'm not too optimistic though, since I fear style sheets
> and the like will soon obsolete any such attempt just like tables
> obsoleted mothra.

The first statement is true, but misses an important point.  The
people who could write that small web browser won't.  For a job
like that you need someone with good aesthetic sense, spartan
coding style, and a strong sense of order.  But the browser has
to render the vast majority of real pages correctly--that is, as
the author intended and not as the (ugly and everchanging) standards
specify.  The problem with HTML in particular is that any idiot can
write it, and most of them do.  If you are in the habit of clicking
"view page source" to see why pages render oddly then you know what
I mean.  If you want a working web browser you have to look at a
vast amount of bad code and try to figure out what the author's intent
was.  If you want to fix links then that code is mostly C code.  If
you want to write your own then the bad code is mostly HTML and
javascript.  I am far too lazy to do either, but I think Andrey
has probably chosen the less painful path.

The second statement is also true, and is one of many reasons why
we are not likely to see the browser equivalent of sam or acme any
time soon.  Who wants to write an elegant solution to a problem which
changes every time a new edition of Internet Explorer appears?
--
John Stalker
Department of Mathematics
Princeton University
(609)258-6469

Re: [9fans] Links for Plan 9

[vic zandy]

how about writing a "browse server"?

the plan 9 client sends a url to the browse server.
the server, running on some other operating system,
calls a real web browser to fetch and render the page.
the server then scrapes the rendered page and returns
it to the client for display.

i can think of dozens of very icky issues -- forms,
pop-ups, finding links, knowing when the page is loaded
(or hung or unavailable), a reload button, cookies, ...

more interesting, i think, is whether their solutions
can be designed independent of the browser -- so that
they can be solved once, perhaps by hideous and extreme
means, but then maintained with light work.

for example, the mechanism for scraping the page might
involve accessing off-screen portions of a bitmap that
have not been displayed (or perhaps even rendered).
i'm sure that someone could corner browser brand X into
coughing the bits.  but is there a general (perhaps
OS-dependent) technique to coherently "scroll" bits
from another program's window?

the incentive for examining this approach is that if
successful it would avoid the well-decried maintenance
hassles of both a native plan 9 browser and an
eternally incomplete port of some other giant browser.

compared to vnc'ing to the browser host, you'd get a
client in a real plan 9 window and process that can
communicate in the usual ways (e.g., plumbing) without
helpers.  it'd be simpler to fire up.  also, perhaps
the browser server could be designed to serve a group
of users, which might be easier on an organization than
supporting vnc sessions for everyone.

alas, this is only a suggestion, not an offer.

Re: [9fans] Links for Plan 9

[mirtchov]

> how about writing a "browse server"?
>
> the plan 9 client sends a url to the browse server.
> the server, running on some other operating system,
> calls a real web browser to fetch and render the page.
> the server then scrapes the rendered page and returns
> it to the client for display.

This has been done already -- there's a script and a plumb rule in the
9fans archives that starts a new tab in Opera's window with the url
being plumbed.  Opera usually runs in a VNC window session.

Re: [9fans] Links for Plan 9

[vic zandy]

no, i propose a native client, not one that runs from
vnc prison.  unlike conventional native clients, web
page rendering would be performed by a remote service.
the payoff for the user is a client that looks and
feels and maybe even performs like a real plan 9
program instead of like a vnc session.

Re: [9fans] Links for Plan 9

[rog]

> no, i propose a native client, not one that runs from
> vnc prison.  unlike conventional native clients, web
> page rendering would be performed by a remote service.
> the payoff for the user is a client that looks and
> feels and maybe even performs like a real plan 9
> program instead of like a vnc session.

so, vnc without the pixels outside the web browser window, then? :-)

there are other problems getting it to interact "seamlessly" though:
it lives in a different file namespace; and what about things like new
browser windows, plumbing from the browser, etc?

not to mention the fact that you need another computer around to do
your rendering for you, not ideal when connected from on a laptop via
your hotel room's IP connection...

and to those that have used charon and found it frustrating, what
were, to you, the most frustrating features (or lack of)? i certainly
have a few of my own, but i'm interested to find out other people's
priorities.

Re: [9fans] Links for Plan 9

[Dan Cross]

rog@vitanuova.com writes:
> and to those that have used charon and found it frustrating, what
> were, to you, the most frustrating features (or lack of)? i certainly
> have a few of my own, but i'm interested to find out other people's
> priorities.

In no particular order, I found it: (a) didn't handle common javascript
errors gracefully.  In most cases, it should probably try to soilder
on.  instead, it'd just give up on ecmascript processing.  (b) when
last I looked at it, I couldn't run it in its own window, independent
of wm/wm anymore.  (c) it loaded pages slowly, if at all.  I often
found myself crossing my fingers that any given page would load.  (d)
often times it rendered things incorrectly.  The javascript problems
were my biggest frustrations, as I recalled.  Too many things depend on
leniency in interpreting javascript, and charon was too strict.  I gave
up on it as a browser long ago.

	- Dan C.

Re: [9fans] Links for Plan 9

[C H Forsyth]

>>Too many things depend on
>>leniency in interpreting javascript, and charon was too strict.

i suspect many of those pages are now fixed,
because Internet Explorer is now by no means lax
or particularly lenient.
On the other hand, Opera gets messed up by some current
javascript, so it's not all plain sailing.

Re: [9fans] Links for Plan 9

[Adrian Tritschler]

rog@vitanuova.com wrote:

..snip..

> and to those that have used charon and found it frustrating, what
> were, to you, the most frustrating features (or lack of)? i certainly
> have a few of my own, but i'm interested to find out other people's
> priorities.

Problems I experienced:

My complete inability to get Inferno and Charon to install and run.  A 
seemingly endless amount of fiddling, rebooting to Windows to download 
files, reboot to plan9 to try installs (a mandatory authenticating web 
proxy meant I can't get anything from the plan9 machine).  Never did get 
Inferno running on plan9, I did have it (Inferno, not Charon) running on 
WinXP at some point, but by then I'd lost interest.

maybe I'll try links!

---------------------------------------------------------------
Adrian Tritschler                          mailto:ajft@ajft.org
Latitude 38°S, Longitude 145°E, Altitude 50m,      Shoe size 44
---------------------------------------------------------------

Re: [9fans] Links for Plan 9

[okamoto]

I recompiled links with Dave's new ape library (select() problem?)
just before (of course including Gorka's plumb, thank you very much Gorka!),
however, it still has select() problem.  And then, it takes lo---ng time to
get http.nytimes.com or www.cnn.com.   Isn't Geoff's work is included?

Kenji

Re: [9fans] Links for Plan 9

[mirtchov]

> Isn't Geoff's work is included?

No, it isn't.

I have also gotten links to render using Plan 9 fonts, which speeds up
things significantly (rendering slashdot in 3 secs vs 15 secs
with Links' bitmap fonts).  I need to figure out proper font sizing and
then you can have it:

	http://pages.cpsc.ucalgary.ca/~mirtchov/screenshots/links-p9fonts.png

andrey

Re: [9fans] Links for Plan 9

[okamoto]

> I have also gotten links to render using Plan 9 fonts, which speeds up
> things significantly (rendering slashdot in 3 secs vs 15 secs
> with Links' bitmap fonts).  I need to figure out proper font sizing and
> then you can have it:

Slashdot can be gotten about 3 secs here...

By the way, the Links goes to stalled when we made Links alone for a
long time?

Kenji

Re: [9fans] Links for Plan 9

[mirtchov]

> By the way, the Links goes to stalled when we made Links alone for a
> long time?

I leave it on overnight without any problems...

Re: [9fans] Links for Plan 9

[John Stalker]

> There are several things unfinished, most notably JavaScript support (turned
> off due to problems with sites like google) and a bug with select()
> descriptors which causes sites like cnn.com, nytimes (every time) and
> theregister (on occasion) to fail loading.
>
> The links code is horrendous, with most of the comments written in Czech
> (funny, in the few places where I understand the comments they seem to be
> horrendous too) with the name Jim Beam interspersed just so foreign readers
> would know what's hapening.

Not that it's really relevant, but the Jim Beam comments are limited to
one file: jsint.c.  Perhaps it's not entirely coincidental that the
biggest outstanding problem is with the javascript interpreter.

--
John Stalker
Department of Mathematics
Princeton University
(609)258-6469