w3 under development or not?

w3 under development or not?

[Joe Corneli]

Anyone working on w3 stuff?  Like

* not causing emacs to hang while looking for a server

Joe

Re: w3 under development or not?

[Björn Lindström]

Joe Corneli <jcorneli@math.utexas.edu> writes:

> Anyone working on w3 stuff?

That's a very good question.

> Like
>
> * not causing emacs to hang while looking for a server

That's a problem with Emacs not being multi-threaded, rather than
w3. You can expect this kind of behaviour to disappear some time after
Emacs has finally been ported to Guile.

-- 
Björn Lindström <bkhl@elektrubadur.se>
http://bkhl.elektrubadur.se/
ICQ: 82945879

Re: w3 under development or not?

[Stefan Monnier]

> That's a problem with Emacs not being multi-threaded, rather than
> w3. You can expect this kind of behaviour to disappear some time after
> Emacs has finally been ported to Guile.

As far as I know, the bug he's referring to is that a piece of elisp
is busy looping waiting for an async event to come up, but the event
never comes (maybe because it passed already).  I.e. it's just the
same as if you had threads.


        Stefan

Re: w3 under development or not?

[A. L. Meyers]

IMHO emacs-w3m is much better and really works.  However, purists will
rightly observe that w3m is not written in elisp.

Cheers!

Lucien
-- 
If you receive this by error, please delete it and inform the sender.
Key fingerprint: 8994 CFDD 9C60 C978 A488 689F 5DCC EFC4 4D36 580D
To Big Brother Echelon from "spook":
opus dei fissionable Kandahar cracking Honduras cryptographic BATF Iraq 

Re: w3 under development or not?

[Stefan Monnier]

> I think this discussion refers to the function url-retrieve-synchronously,
> where the following piece of code can be found:

> 	;; Quoth monnier:
> 	;; It turns out that the problem seems to be that the (sit-for
> 	;; 0.1) below doesn't actually process the data: instead it
> 	;; returns immediately because there is keyboard input
> 	;; waiting, so we end up spinning endlessly waiting for the
> 	;; process to finish while not letting it finish.

> 	;; (sit-for 0.1)
> 	(sleep-for 0.1))

> Your change did not resolve the problem completely, so I changed it to
> (sit-for 0 250) followed by (discard-input). I am not sure if it is a

That sounds interesting.  Are you saying that you not only see hangs with
my code but you also see hangs with the old code (in the only case I've
heard of where someone had trouble with my version of the above code,
that smeone also said that reverting my change fixed his problem).
Maybe instead of discard-input we could do something like

      (while (input-pending-p)
        (push (read-event) w3-temp-read-events))

and then push those events back onto unread-command-events when the whole
thing is done ?

Still, it seems the problem is revealing a bug in the underlying C code,
and it would be even better to fix that one.


        Stefan

Re: w3 under development or not?

[Joe Corneli]

> By the way: What is the recommended way to suggest improvements to
> Emacs/W3? I have some more work-arounds in my .emacs that I would
> like to see in the standard.

This gets back to the first part of my question.  There is a w3
mailing list, but it is apparently very low traffic and you probably
already know about it, but anyway here is the link...

http://mail.gnu.org/mailman/listinfo/w3-dev

One might hope that this was the right place to suggest improvements;
anyway, I found this mailing list before asking about w3 here. However
it is so low traffic that I worry that it may not be the right place.

???

Joe

Re: w3 under development or not?

[Friedrich Dominicus]

bkhl@elektrubadur.se (Björn Lindström) writes:

> That's a problem with Emacs not being multi-threaded, rather than
> w3. You can expect this kind of behaviour to disappear some time after
> Emacs has finally been ported to Guile.
you expect an Emacs guile? See you in the next century :)

It's varpoware at it's worst. 

Regards
Friedrich

Re: w3 under development or not?

[Shane]

bkhl@elektrubadur.se (Björn Lindström) writes:

> Joe Corneli <jcorneli@math.utexas.edu> writes:
> 
> > Anyone working on w3 stuff?
> 
> That's a very good question.
> 
> > Like
> >
> > * not causing emacs to hang while looking for a server
> 
> That's a problem with Emacs not being multi-threaded, rather than
> w3. You can expect this kind of behaviour to disappear some time after
> Emacs has finally been ported to Guile.

I know this is slightly off-topic, but I have the same problem when I
download my email under GNUs. It hangs for a long time while 50 of the
latest "Microsoft Internet Security patches" or other crap are
retrieved. During this wait, I would prefer to be working in a another
Emacs window, but Emacs won't accept any input.

As suggested, is it a "design feature" of Emacs rather than
configuration problem?

> 
> -- 
> Björn Lindström <bkhl@elektrubadur.se>
> http://bkhl.elektrubadur.se/
> ICQ: 82945879

Re: w3 under development or not?

[Peter Lee]

>>>> Shane  writes:

    Shane> I know this is slightly off-topic, but I have the same
    Shane> problem when I download my email under GNUs. It hangs for a
    Shane> long time while 50 of the latest "Microsoft Internet
    Shane> Security patches" or other crap are retrieved. During this
    Shane> wait, I would prefer to be working in a another Emacs
    Shane> window, but Emacs won't accept any input.

I was getting about 50-80 of those a day myself for a while... I wrote
a perl script to parse a .spam file containing one regex per line.  If
any of the regexes in the .spam file match the author or subject in
the mail on the server, it's deleted on the server.  I mainly just
focused on the service packs as they took so long to download.  I'm a
perl novice, so I'm sure this code could be reduced to about 10 lines
by someone that knows the language better.

Here's a snippet from my .spam:

critical.*update
critical.*upgrade
customer.*bulletin
public.*assistanc
internet.*system
security.*pack
security.*update
security.*patch
delivery.*system
microsoft

And the script (I called pop-spam.pl) I just call with no arg for
preview: 'perl pop-spam.pl', and 'perl pop-spam.pl delete' when I'm
certain of my regexes.


#!e:/perl/bin/perl
# 
use strict;
use Mail::POP3Client;

open SPAM, "<.spam"
    or die "Failed to open .spam: $!";

my ($parm) = @ARGV;

my @spam = <SPAM>;
chomp @spam;

my $pop = new Mail::POP3Client(USER      => "yourusername",               
                               PASSWORD  => "yourpasswd",            
                               HOST      => "your.pop3.server",
                               AUTH_MODE => 'PASS',
                               TIMEOUT   => 30,
                               LOCALADDR => undef,
                               DEBUG     => 0);

for (my $i = 1; $i <= $pop->Count(); $i++) 
    {
    my @lines = $pop->Head($i);
    my $subj;
    my $from;

    foreach my $ln (@lines)
        {
        if ($ln =~ /^From:/i)
            {
            $from = $ln;
            }
        elsif ($ln =~ /^Subject:/i)
            {
            $subj = $ln;
            }
        }
    
    my ($msg, $size) = split ' ', $pop->List($i);
    my $deleted = 0;

    print "($msg) $size bytes\n";
    print $from, "\n";
    print $subj, "\n";

    foreach my $s (@spam)
        {
        if ($from =~ /$s/i)
            {
            print "-DELETE FROM ($s)\n";
            if ($parm =~ /delete/i)
                {
                $pop->Delete( $i );
                $deleted = 1;
                }
            last;
            }
        elsif ($subj =~ /$s/i)
            {
            print "-DELETE SUBJECT ($s)\n";
            if ($parm =~ /delete/i)
                {
                $pop->Delete( $i );
                $deleted = 1;
                }
            last;
            }
        }
    
    if ($deleted == 0)
        {
        print "-OK\n";
        }

    print "\n";
    }

$pop->Close();
close SPAM;

Re: w3 under development or not?

[Ted Zlatanov]

On Fri, 14 Nov 2003, pete_lee@swbell.net wrote:

> I was getting about 50-80 of those a day myself for a while... I
> wrote a perl script to parse a .spam file containing one regex per
> line.  If any of the regexes in the .spam file match the author or
> subject in the mail on the server, it's deleted on the server.  I
> mainly just focused on the service packs as they took so long to
> download.  I'm a perl novice, so I'm sure this code could be reduced
> to about 10 lines by someone that knows the language better.
> 
> Here's a snippet from my .spam:
> 
> critical.*update
> critical.*upgrade
> customer.*bulletin
> public.*assistanc
> internet.*system
> security.*pack
> security.*update
> security.*patch
> delivery.*system
> microsoft
> 
> And the script (I called pop-spam.pl) I just call with no arg for
> preview: 'perl pop-spam.pl', and 'perl pop-spam.pl delete' when I'm
> certain of my regexes.

This is a really dangerous recommendation.  If you just want to
eliminate executable attachments, which you seem to be trying to do,
use something like this Procmail recipe:

:0 H
* ^Content-Type: multipart
# check body for executable filenames
{
  :0 B
  * name=.*\.(exe|bat|pif|com|lnk|scr|vbs)(")?(\ *|\t*)$
    {
     :0
     /wherever/spam
    }
}

or install a program designed for that purpose, e.g. SpamAssassin.
Matching regular expressions against the subject is risky - are you
sure you will never want to find out about security packages, or
receive mail from people at Microsoft?

It's even riskier to delete message you haven't seen, but that's your
choice.

Ted

Re: w3 under development or not?

[Peter Lee]

>>>> Ted Zlatanov writes:

    Ted> This is a really dangerous recommendation.  If you just want
    Ted> to eliminate executable attachments, which you seem to be
    Ted> trying to do, use something like this Procmail recipe:

    Ted> or install a program designed for that purpose,
    Ted> e.g. SpamAssassin.  

Are those tools able to delete on the server?  If so, and they work on
windows, and are fairly simple to set up, I'll have a look.

    Ted> Matching regular expressions against the subject is risky -

I agree.  I wouldn't use it on an important account.

    Ted> are you sure you will never want to find out about security
    Ted> packages, or receive mail from people at Microsoft?

On this particular account, I'm positive.

But your point is well taken.

Re: w3 under development or not?

[Harry Putnam]

Peter Lee <pete_lee@swbell.net> writes:

> critical.*update
> critical.*upgrade
> customer.*bulletin
> public.*assistanc
> internet.*system
> security.*pack
> security.*update
> security.*patch
> delivery.*system
> microsoft
>
> And the script (I called pop-spam.pl) I just call with no arg for
> preview: 'perl pop-spam.pl', and 'perl pop-spam.pl delete' when I'm
> certain of my regexes.

I've found a fairly simple procmailrc recipe that gets %99 of thos ms
thingies:

  :0HB:
  * ^Content-Type:.*(audio/x\-|x\-msdownload)
  viral_spam1.in
Nearly all are either audio/x-something or -msdownload
And since I get no other audio attachments het above is all I need.

Re: w3 under development or not?

[Ted Zlatanov]

On Fri, 14 Nov 2003, pete_lee@swbell.net wrote:

>>>>> Ted Zlatanov writes:
> 
>     Ted> This is a really dangerous recommendation.  If you just
>     Ted> want to eliminate executable attachments, which you seem to
>     Ted> be trying to do, use something like this Procmail recipe:
> 
>     Ted> or install a program designed for that purpose,
>     Ted> e.g. SpamAssassin.  
> 
> Are those tools able to delete on the server?  If so, and they work
> on windows, and are fairly simple to set up, I'll have a look.

There are many, many such tools.  For Unix mail servers the options
are plentiful and a simple web search will turn them up.  I don't use
Windows mail servers but I'm sure similar tools exist for them.

>     Ted> are you sure you will never want to find out about security
>     Ted> packages, or receive mail from people at Microsoft?
> 
> On this particular account, I'm positive.

I think that's pretty dangerous, but it's certainly your prerogative
to block mail from anyone.  Recommending that to a newsgroup,
however, merits a little more explanation :)  Generally, solutions
that are manually maintained tend to fall behind the spam - you're
trying to block yesterday's spam.  It may work, but it's a lot of
effort.

Ted

Re: w3 under development or not?

[Peter Lee]

>>>> Ted Zlatanov writes:

    Ted> There are many, many such tools.  For Unix mail servers the
    Ted> options are plentiful and a simple web search will turn them
    Ted> up.  I don't use Windows mail servers but I'm sure similar
    Ted> tools exist for them.

After a brief look at Procmail and SpamAssassin, I don't see how they
solve the problem.  Installing either locally still results in the
mail being pulled down my small pipe from swbell.  The only benefit I
see to them is if your mail client either:
a) doesn't have spam support
b) retrieves mail in a synchronous fashion

Neither a or b apply to Outlook.  Unfortunately b does apply to Gnus
which is my preferred client.

So I could use one of those to prevent Gnus from locking up while
retrieving 20-80 of those patches, but they will still have to come
down my connection, potentially when I'm doing something important
like playing Half-life or Quake.

I suppose I could lobby for swbell to install SpamAssassin.  But then
I'm dependent on someone else setting up rules that could potentially
cost me a valid email now and again.  They'd never do it anyway.

What would be cool is if (in Gnus) you could download mail headers
only first... run through spam processor... then only download valid
mail, and reprocess again including body.  The spam would still show
up in spam group, possibly annotated to show that only the headers
have been retrieved on some (hopefully most).  If you moved it to a
ham group it would then fetch the mail, otherwise it would be expired
and on expiry be deleted from the server the next time you asked for
mail.

It would be slower, but not as slow as downloading 20-80 of those
patches.

Re: w3 under development or not?

[Juri Linkov]

Peter Lee <pete_lee@swbell.net> writes:
> I suppose I could lobby for swbell to install SpamAssassin.  But then
> I'm dependent on someone else setting up rules that could potentially
> cost me a valid email now and again.  They'd never do it anyway.

I highly recommend the server-side spam filtering, e.g. SpamAssassin.
If your provider don't have it, you could find a provider that has
such service.  If those M$ security patches are removed on server,
then you have no more problems with hanging Gnus.  However, currently
SpamAssassin scores for Micro$oft executable programs are not
well-balanced, so I suggest you to add the following line to the
user_prefs file:

score MICROSOFT_EXECUTABLE 4.5

This works so well for me.  Once a month I quickly browse spam headers
collected in the spam folder on the server (to check whether all
they are spam) and delete them.  I am so happy now to not receive
anymore all those spam letters about Sexy Lisp (I'm not sure is it
Emacs Lisp what they mean :-)

-- 
http://www.jurta.org/emacs/

Re: w3 under development or not?

[Ted Zlatanov]

On Fri, 14 Nov 2003, pete_lee@swbell.net wrote:

>>>>> Ted Zlatanov writes:
> 
>     Ted> There are many, many such tools.  For Unix mail servers the
>     Ted> options are plentiful and a simple web search will turn
>     Ted> them up.  I don't use Windows mail servers but I'm sure
>     Ted> similar tools exist for them.
> 
> After a brief look at Procmail and SpamAssassin, I don't see how
> they solve the problem.  Installing either locally still results in
> the mail being pulled down my small pipe from swbell.  

You asked for server-side options.  Procmail and SA are best
installed on the server if you want to cut down the download times.

> I suppose I could lobby for swbell to install SpamAssassin.  But
> then I'm dependent on someone else setting up rules that could
> potentially cost me a valid email now and again.  

With any reasonable ISP that provides you a shell account, you can
set up your personal rules in ~/.spamassassin/user_prefs

I use pair.com but there are many others.

> What would be cool is if (in Gnus) you could download mail headers
> only first... run through spam processor... then only download valid
> mail, and reprocess again including body.  The spam would still show
> up in spam group, possibly annotated to show that only the headers
> have been retrieved on some (hopefully most).  If you moved it to a
> ham group it would then fetch the mail, otherwise it would be
> expired and on expiry be deleted from the server the next time you
> asked for mail.

You can actually do all that and more with spam.el, a package for Gnus
which I maintain.  Check out the CVS version of Gnus and look in the
manual (Oort Gnus comes with spam.el also, but many bugs have been
fixed meanwhile).  spam.el is sort of a framework, and inside it are
various flexible spam checks you can enable depending on your needs.
I would still do server-side filtering in addition to client-side
filtering, but that's your choice and moving away from swbell is
certainly going to be a hassle for you.

To only get mail headers, use IMAP.  There's more to it, I suggest you
read the Gnus manual in the spam.el section, but what you describe is
definitely possible - and you should continue the discussion on the
gnu.emacs.gnus newsgroup if you have any questions.

Ted

Re: w3 under development or not?

[Alan Mackenzie]

Peter Lee <pete_lee@swbell.net> wrote on Fri, 14 Nov 2003 22:39:24 GMT:
>>>>> Ted Zlatanov writes:

>     Ted> There are many, many such tools.  For Unix mail servers the
>     Ted> options are plentiful and a simple web search will turn them
>     Ted> up.  I don't use Windows mail servers but I'm sure similar
>     Ted> tools exist for them.

> After a brief look at Procmail and SpamAssassin, I don't see how they
> solve the problem.  Installing either locally still results in the mail
> being pulled down my small pipe from swbell.  The only benefit I see to
> them is if your mail client either: a) doesn't have spam support b)
> retrieves mail in a synchronous fashion

My ISP uses SpamAssassin.  I can rlogin to my ISP's mailserver and run a
mail client there (pine, as it happens).  I can use this to mark and
delete spam _before_ I download it through my modem.

[ .... ]

> I suppose I could lobby for swbell to install SpamAssassin.  But then
> I'm dependent on someone else setting up rules that could potentially
> cost me a valid email now and again.  They'd never do it anyway.

Why not ask them?  Are they happy having their modem lines needlessly
tied up whilst you download 20 or 80 "Microsoft last security patch"es?
Have you got a remote login facility at swbell?

> What would be cool is if (in Gnus) you could download mail headers
> only first... run through spam processor... then only download valid
> mail, and reprocess again including body.  The spam would still show
> up in spam group, possibly annotated to show that only the headers
> have been retrieved on some (hopefully most).  If you moved it to a
> ham group it would then fetch the mail, otherwise it would be expired
> and on expiry be deleted from the server the next time you asked for
> mail.

This is effectively what I do with an rlogin.

> It would be slower, but not as slow as downloading 20-80 of those
> patches.

-- 
Alan Mackenzie (Munich, Germany)
Email: aacm@muuc.dee; to decode, wherever there is a repeated letter
(like "aa"), remove half of them (leaving, say, "a").