🚀 go-pugleaf

On Wednesday, in article <817661125snz@keris.demon.co.uk>
     chris@keris.demon.co.uk "Chris Croughton" wrote:

CC> Some time ago someone (John Washington?) said that they had a version of
CC> the SNEWS ADDGROUP program that generated names which bore some
CC> resemblance to the newsgroup, rather than the usual time-based ones.

In case anyone else wants, I'll post it here, also just in case I've
forgotten to mail it to Malcolm.

If you dig around in newsgroup files at all, I simply can't
imagine being without this mod.

When is the next Snews release, Malcolm?  I've been holding back
on doing further mods until 1.31, in particular because I hope
you'll do more restructuring to reduce the large chunks of repeated
code.

Changes are to addgroup.c, addition of new files nicename.c/.h,
and the obvious changes to makefile (exercise for rdr).  Sorry I
can't give diffs, or line numbers, I've made too many other changes;
I'll quote a little context instead.

I've expanded tabs to 4 spaces.  (Tabs?  Just say 'no').

-------------------------------------
Mods to addgroup.c:
#include "locking.h"

#ifdef JW_NICE_FILENAMES
#include "nicename.h"
#endif

-------------------------------------
    #ifndef JW_NICE_FILENAMES
    time_t          t;
    #endif
-------------------------------------
                added = added2 = FALSE;
                if (active_file != NULL) {

                    #ifdef JW_COMMENT
                    //> Strange logic, has to be true, see above.
                    #endif

                    #ifndef JW_NICE_FILENAMES
                    /* delay ensures names are different - ahem */
#ifdef __OS2__
                    os2_delay(1000);
#else
#ifdef __TURBOC__
                    delay(1000);
#else
                    goal = 1000 + clock();
                    while (goal > clock());
#endif
#endif
                    sprintf(txt_name, "%ld", time(&t));
                    if (strlen(txt_name) > 8) {
                        strcpy(txt_name, txt_name + (strlen(txt_name) - 8));
                    }
                    #else
                    {
                        // Create a 'sensible' name
                        nice_new_newsgroup_name (argv[i], txt_name, sizeof txt_name);
                        printf ("\n%s %s\n", argv [i], txt_name);   // Print the filename chosen, just for interest.
                    }
                    #endif  // else def JW_NICE_FILENAMES
-------------------------------------
                        fprintf(stderr, "addgroup: newsbase name collision - try again\n");
                        fclose(new_active_file);

                        #ifdef JW_NICE_FILENAMES
                        // *** BUG *** BUG *** BUG ***
                        // I don't think this should be here even for the old code.  
                        // I certainly don't want it with my new code.
                        #else
                        unlink(buf);
                        #endif

                        exit(1);
-------------------------------------
New file nicename.c:
// nicename.c

// Nice filenames in ~\spool\news\newsbase.
// This module is only used by addgroup.exe (and mail2snz.exe).

#include "defs.h"

#ifdef JW_NICE_FILENAMES

#include <ctype.h>  // for isdigit
#include <io.h>
#include <stdio.h>
#include <string.h>

#include "nicename.h"


void nice_new_newsgroup_name (const char* ngn, char* txt_name, int txt_name_len)
{
    const char* lastbit = ngn + 1;
    int j;
    
    if (txt_name_len < 9)
    {
        txt_name [0] = '\0';
        return;
    }

    txt_name [0] = ngn [0];
    for (j = 1; j < 7; j++)     // j points at next unset ch in txt_name.
    {
        // Find the next '.'
        for (;; ngn++)
        {
            if (*ngn == '.')
            {
                ++ngn;
                break;
            }
            if (*ngn == '\0')
                break;
        }
        if (*ngn == '\0' 
            || *ngn == '.')     // (firewall in case ".." in n/grp)
        {
            break;
        }
        txt_name [j] = *ngn;
        lastbit = ngn + 1;
    }
    if (j > 1)
    {           // Make the name prettier, by adding underscore
        txt_name [j] = txt_name [j-1];
        txt_name [j-1] = '_';
        ++j;
    }
    if (j < 8)
        strncpy (txt_name + j, lastbit, 8 - j);
    txt_name [8] = '\0';

    // txt_name[] now holds a name up to 8 chars, a candidate
    // for the new file to hold this newsgroup.  Collisions
    // are possible, so allow the final character to be replaced
    // with a digit.  In practice this should give sufficient
    // variety, but something smarter could replace more than
    // the last char.

    j = '0';    // Prepare to try replacing final ch with digit
    for (;;)
    {
        char buf [256];
        sprintf(buf, "%snewsbase\\%s", my_stuff.news_dir, txt_name);
        if (access(buf, 0) != 0)
            break;      // Success, found a name that doesn't exist.
        if (!isdigit (j) 
            || txt_name [0] == '\0')  // (imposs, but test anyway)
            break;      // Failure, let the caller sort it out.
        txt_name [strlen (txt_name) - 1] = j;
        j++;
    }
} // nice_new_newsgroup_name

#endif  // def JW_NICE_FILENAMES

// EOF
-------------------------------------
New file nicename.h:
// nicename.h

// Nice filenames in ~\spool\news\newsbase.
// This module is only used by addgroup.exe.

#ifdef JW_NICE_FILENAMES

void nice_new_newsgroup_name (const char* ngn, char* txt_name, int txt_name_len);

#else
    #error nicename.h shouldn't be included.
#endif

// EOF
-------------------------------------  end of mods & additions

In article <817747310snz@wash.demon.co.uk>
           john@wash.demon.co.uk "John Washington" writes:

>Someone could write a small test program, using nicename.c, to
>generate a list of filenames for the whole of the current
>'active.dis'.  The list would be output to another file.
>That could then be scanned, for non-uniqueness, anomalies,
>what-have-you.

Well, I just wrote that program, and with the aid of a couple of
programs (sort and a couple of command-line awk scripts) ran it on the
active list (as of yesterday) and got some stats:

# newsgroups:   13956
# duplications:  2128 (that's duplicate names your code generates)
max. dupl.:        24 (max. same name generated)

That's without your code to see if the file exists already, of course.
Most of the duplications are on newsgroups ending in .misc, .general and
.test, of course, but there were a large number ending .bug as well
<g>.  So in the worst case, your code would need to generate 24 unique
names - I think your code just replaces the last character with a digit,
which won't be enough.

One suggestion could be to replace the underscore with a digit as well,
or hyphen (there aren't many other characters legal in DOS filenames -
quote, hash, bang?)...

On my active file, with 28 newsgroups, there were no duplications...

If you're interested, the awk script I used (on a sorted file with each
line having the generated filename, scace and then the newsgroup name)
was as follows (de-obfuscated - the commandline version is shorter <g>):

{
    if (prev == $1 || last)
    {
        if (!last) print blank;
        print prevline;
    }
    last = (prev==$1);
    prev = $1;
    prevline = $0;
}

***********************************************************************
*  chris@keris.demon.co.uk      *                                     *
*  chriscr@cix.compulink.co.uk  *  FIAWOL (Filking Is A Way Of Life)  *
*  100014.3217@compuserve.com   *                                     *
***********************************************************************

In article <817747310snz@wash.demon.co.uk>
   John Washington <john@wash.demon.co.uk> writes:
> 
> Someone could write a small test program, using nicename.c, to
> generate a list of filenames for the whole of the current
> 'active.dis'.  The list would be output to another file.
> That could then be scanned, for non-uniqueness, anomalies,
> what-have-you.

From this, a program which scanned ACTIVE and renamed existing newsgroup 
files didn't ought to be too hard. Not, you understand, that I have any 
interest in this as a non-Snews user....

I must admit that I've been thinking of doing somthing along these lines 
for some time, so maybe I'll have a look at your code in a quiet moment 
(are there any?); I'll let this group know in the unlikely event that 
something comes of it.

Mark.
-- 
int MeaningOfLife (void) { return 42; }   // mcr@holly.demon.co.uk
function MeaningOfLife :integer; begin MeaningOfLife := 6 * 9 end;

On Wednesday, in article <817661125snz@keris.demon.co.uk>
     chris@keris.demon.co.uk "Chris Croughton" wrote:

CC> Some time ago someone (John Washington?) said that they had a version of
CC> the SNEWS ADDGROUP program that generated names which bore some
CC> resemblance to the newsgroup,

A further thought, having posted the code I use...

Someone could write a small test program, using nicename.c, to
generate a list of filenames for the whole of the current
'active.dis'.  The list would be output to another file.
That could then be scanned, for non-uniqueness, anomalies,
what-have-you.

The function has worked well for me, for over a year, but
I'm not very adventurous in my Usenet travels.  Such a test
might give greater confidence in the code.

--
John Washington

John Washington (john@wash.demon.co.uk) wrote:
> On Wednesday, in article <817661125snz@keris.demon.co.uk>
>      chris@keris.demon.co.uk "Chris Croughton" wrote:

> CC> Some time ago someone (John Washington?) said that they had a version of
> CC> the SNEWS ADDGROUP program that generated names which bore some
> CC> resemblance to the newsgroup, rather than the usual time-based ones.

> In case anyone else wants, I'll post it here, also just in case I've
> forgotten to mail it to Malcolm.

> If you dig around in newsgroup files at all, I simply can't
> imagine being without this mod.

> When is the next Snews release, Malcolm?  I've been holding back
> on doing further mods until 1.31, in particular because I hope
> you'll do more restructuring to reduce the large chunks of repeated
> code.

No specific plans as the current version appears to be reasonably
stable.

-- 
Malcolm S. Muir
Sunderland
England

On Friday, in article <817783466.11023@thokk.demon.co.uk>
     Malcolm@demon.net "Malcolm Muir" wrote:

MM> No specific plans as the current version appears to be reasonably
MM> stable.

Hmmm, time to fix that, then.  ;-)

--
John Washington

On Thursday, in article <817773091snz@keris.demon.co.uk>
     chris@keris.demon.co.uk "Chris Croughton" wrote:

CC> >Someone could write a small test program, using nicename.c, to
..
CC> Well, I just wrote that program, and with the aid of a couple of
CC> programs (sort and a couple of command-line awk scripts) ran it on the
CC> active list (as of yesterday) and got some stats:
CC> 
CC> # newsgroups:   13956
CC> # duplications:  2128 (that's duplicate names your code generates)
CC> max. dupl.:        24 (max. same name generated)

What a star!  Thanks Chris!

..
CC> <g>.  So in the worst case, your code would need to generate 24 unique
CC> names - I think your code just replaces the last character with a digit,
CC> which won't be enough.

Good point.

CC> One suggestion could be to replace the underscore with a digit as well,
CC> or hyphen (there aren't many other characters legal in DOS filenames -
CC> quote, hash, bang?)...

How about (since this will never happen in practice anyway, well,
hardly ever <g>) just working backwards, which, with less than
99 clashes means just a 2nd digit?

However, some names are less than 8 chars.  I wonder if extending
such names would be worthwhile?

..
CC> If you're interested, the awk script I used (on a sorted file with each
..

A succint star, too.  Good stuff, this Awk thingie.

--
John Washington

On Thursday, in article <817745608snz@wash.demon.co.uk>
     john@wash.demon.co.uk "John Washington" wrote:

JW> On Wednesday, in article <817661125snz@keris.demon.co.uk>
JW>      chris@keris.demon.co.uk "Chris Croughton" wrote:
JW> 
JW> CC> Some time ago someone (John Washington?) said that they had a version of
JW> CC> the SNEWS ADDGROUP program that generated names which bore some
JW> CC> resemblance to the newsgroup, rather than the usual time-based ones.
..

I forgot to mention, when posting that code, that I think there
is a relevant problem in the locking code used by Snews and
its utilities.  I believe (other opinions, pls) that 7 char
filenames cause problems.  See locking.c, function mlock().

If you agree there's a bug, and you don't fix and recompile all the code
that uses mlock(), I imagine you would wish to modify nicename.c so as
to avoid generating 7 char filenames, e.g. by adding a trailing '_'.

--
John Washington

In article <817904288snz@wash.demon.co.uk>
           john@wash.demon.co.uk "John Washington" writes:

>How about (since this will never happen in practice anyway, well,
>hardly ever <g>) just working backwards, which, with less than
>99 clashes means just a 2nd digit?

I thought of that.  Of course, the more characters you go back, the more
chance of a clash <g>.  Now, if this were a decent filesystem I would
just vary the case (in fact, that was my first idea) - but then if it
were a decent system we'd have more than 8+3 characters anyway...

>However, some names are less than 8 chars.  I wonder if extending
>such names would be worthwhile?

Certainly it would in the case of things like the test newsgroups
(although I don't know why anyone would want to subscribe to more than
one of those anyway!).

That's why I thought of replacing the underscore - no change in meaning,
in general, and it's not dropping any significant characters...

>A succint star, too.  Good stuff, this Awk thingie.

It produces convoluted command lines, though - I'm just glad 4DOS has a
255 character limit not 127!  I'm using awk quite a lot; Perl looks more
powerful, but the syntax seems more clumsy to me...

***********************************************************************
*  chris@keris.demon.co.uk      *                                     *
*  chriscr@cix.compulink.co.uk  *  FIAWOL (Filking Is A Way Of Life)  *
*  100014.3217@compuserve.com   *                                     *
***********************************************************************

In article <817783466.11023@thokk.demon.co.uk>
           Malcolm@demon.net "Malcolm Muir" writes:

> John Washington (john@wash.demon.co.uk) wrote:
> 
> > When is the next Snews release, Malcolm?  I've been holding back
> > on doing further mods until 1.31, in particular because I hope
> > you'll do more restructuring to reduce the large chunks of repeated
> > code.
> 
> No specific plans as the current version appears to be reasonably
> stable.
> 

There des seem to be a teeny-weeny bug, in that after a search, the words
do not always turn green.  I wasn't going to report this until I found a
way to repeat it, because at the moment it seems pretty random.

-- 
Tom Wheeley, <holyhorns>
Oh, did I mention that I'm going to see Bottom Live next Thursday :-))
* TQ 1.0 * Stolen Sigs!
"Why don't we just give Bill Gates all our money now and get it over with?"
"Pride"       -- Doonesbury
[cspgdl@dcs.hull.ac.uk]

Tom Wheeley (tomw@tsys.demon.co.uk) wrote:
> In article <817783466.11023@thokk.demon.co.uk>
>            Malcolm@demon.net "Malcolm Muir" writes:

> > John Washington (john@wash.demon.co.uk) wrote:
> > 
> > > When is the next Snews release, Malcolm?  I've been holding back
> > > on doing further mods until 1.31, in particular because I hope
> > > you'll do more restructuring to reduce the large chunks of repeated
> > > code.
> > 
> > No specific plans as the current version appears to be reasonably
> > stable.
> > 

> There des seem to be a teeny-weeny bug, in that after a search, the words
> do not always turn green.  I wasn't going to report this until I found a
> way to repeat it, because at the moment it seems pretty random.

I have also received an update (thanks) that retains the highlight
when the screen is scrolled.

-- 
Malcolm S. Muir
Sunderland
England

On Sunday, in article <818020214snz@lloyney.demon.co.uk>
     rob@lloyney.demon.co.followup-to "Rob Coles" wrote:

RC> Thanks for posting the source John.. I don`t suppose you`d consider also
RC> posting your routines for quoting the initials of the poster when following
RC> up? I seem to remember seeing in your messages.

Would you believe if I said I add them by hand?  :-)

And, warning, some people don't like the way I push
quotes more quickly off the screen.

There's a new module of about 350 lines, including some
changes that would not be to everyone's taste
(e.g. not a friendly "you wrote", but "Rob Coles wrote").
And the "On Sunday" that I'm so proud of, is a bit of
a prat-fall, since it still says that even if you posted
4 months ago.

Plus, there's sizeable hacks to post.c and mail.c to
turn 3 large essentially identical pieces of code into
one, otherwise the 350 lines would have been (hang on,
boot Windoze, double click Calc icon) 1050 lines.

You're welcome to my changes by mail.  I would
immediately add that the most important change of all is
a change to your headers, so it's clear you're not using
Demon's std release.

--
John Washington

In article <818118642snz@wash.demon.co.uk>, John Washington
<john@wash.demon.co.uk> writes

>On Sunday, in article <818020214snz@lloyney.demon.co.uk>
>     rob@lloyney.demon.co.followup-to "Rob Coles" wrote:
>
>RC> Thanks for posting the source John.. I don`t suppose you`d consider also
>RC> posting your routines for quoting the initials of the poster when following
>RC> up? I seem to remember seeing in your messages.
>
>Would you believe if I said I add them by hand?  :-)
>
>And, warning, some people don't like the way I push
>quotes more quickly off the screen.

and some people don't think they're appropriate for Usenet where almost
no clients support them :(  The FIDO readers which support them rely on
seeing leading spaces on the lines involved so as to avoid double
quoting, but this leads to other difficulties when there are spaces at
the start of lines for other reasons.  see news.software.readers passim.

>--
>John Washington

the space on your "-- " has been eaten by something :(

-- 
richard                      Richard Clayton         T U R N P I K E  Ltd
                                                     tel: +44 1306 747747
"Assembly of Japanese bicycle require great peace of mind" quoted in ZAMM

In article <818286975snx@warren.demon.co.uk>
           leveret@warren.demon.co.uk "Nick Leverton" writes:

> A lot of surprising things are legal in DOS filenames.  Space is, for
> instance, though I haven't got a full list to hand !

From section 3.2 of Microsoft MS-DOS Programmer's Reference, ISBN
1-55615-329-5 has:

> Names and extensions can contain letters, digits, and any of the
> characters in the following set:
>     !  #  $  %  ^  &  (  )  - _    {   }   ~

     33 35 36 37 94 38 40 41 45 95 123 125 126
> ...
> Although extended ASCII characters (characters with values greater than
> 127 are also permitted in names, programs should avoid them, since the
> meaning of the extended characters may differ with different code pages.

My memory is that nearly anything goes in practice.

I knocked up a noddy using Borland C 3.1 and found the following extra
characters are acceptable in DOS 6.20
039A'B                   7 06/12/95   22:48
064A@B                   7 06/12/95   22:48
096A`B                   7 06/12/95   22:48
127AB                   7 06/12/95   22:48

I couldn't be bothered working at the interrupt level to factor out any
vagaries of the compiler. Silliness has its limitations!

The following decimal character ranges were unacceptable:
0-31,34,42-44,46,47,59-63,91-93,124
--------
walter briscoe

In article <817773091snz@keris.demon.co.uk>
   Chris Croughton <chris@keris.demon.co.uk> writes:
>One suggestion could be to replace the underscore with a digit as well,
>or hyphen (there aren't many other characters legal in DOS filenames -
>quote, hash, bang?)...

A lot of surprising things are legal in DOS filenames.  Space is, for
instance, though I haven't got a full list to hand !  Command.com's
filename parsing has to be stricter because of the command-line meanings
of |, <, >, but I think almost anything is legal according to DOS.
There used to be an editor floating around that didn't check filenames
for being command.com legal but relied on the DOS call, and it was a
regular support question as to how to remove a DOS file it'd created
with a space in the name.  (The best answer is to use rename with a
wildcard, which again bypasses command.com's checking).

And then there's "EA DATA . SF" :-)

Nick

[ reguarding funny quote lines ]

There are actually several reasons I could quote for *not* using quote
lines of the style User>

1. Its pushes the orginal article out the way quicker.
2. It makes writting code to rewrap an article almost impossible and
   certainly very difficult. Newsreaders are begining to do this now, both
   the beta version of slrn and bnr (my newsreader) do this.
3. It doesn't look right when it itself is quoted.
4. It doesn't match with what son of RFC1036 recommends.. to quote..

   When responding to a previous article, posters SHOULD not assume
   that all readers understand the exact context.  It is common to
   quote some of the previous article to establish context.  This
   SHOULD  be done  by  prefacing  each  quoted line (even if it is
   empty) with the character ">".  This will result in multiple levels
   of ">" when quoted context itself contains quoted context.

5. I don't like it :-)
6. IMHO it is visually displeasing.

I should also note that I personally don't know of any newsreaders that can
actually do this type of quoting, everyone who I've ever seen do it who I've
asked has done it manually.


-- 
Brian Blackmore                         http://www.wonderland.org/~eternal/

John Washington <john@wash.demon.co.uk> wrote:
[ deletion ]
>RC> 
>RC> >--
>RC> >John Washington
>RC> 
>RC> the space on your "-- " has been eaten by something :(

>Yes, the editor I use.  It has many other advantages that
>compensate.  Just think of my posts as having no .sig, but
>their bodies ending in a couple of dashes and my name.

>Do you have a problem with that?

>I have considered fixing it by amending Snews to search
>backwards from the end of my posts for a "--" within the
>last 4 lines, but it's not come high enough up my wish
>list yet.  Besides, since I often review outgoing mail
>later, some would get eaten again.  I've also considered
>a simple utility to scan mail before connecting, but I'm
>reluctant to have the additional delay (as well as the
>additional effort programming it).  However, if you can
>explain to me why it's so important, of course I can
>review my priorities.

The purpose of fixing the signature separation line is quite simple..

1. Its the standard you should keep to it, its also what son of RFC1036
   says which although not being a standard is worth following.
2. Several newsreaders allow you to treat the signature specially so that
   long signatures don't cause articles to page unnecessarily.

Of course what really needs fixing is your editor :-)

-- 
Brian Blackmore                         http://www.wonderland.org/~eternal/

On Wednesday, in article <Ig$bdZArOYxwEACz@turnpike.com>
     richard@turnpike.com "Richard Clayton" wrote:

RC> In article <818118642snz@wash.demon.co.uk>, John Washington
RC> <john@wash.demon.co.uk> writes
RC> 
RC> >On Sunday, in article <818020214snz@lloyney.demon.co.uk>
RC> >     rob@lloyney.demon.co.followup-to "Rob Coles" wrote:
RC> >
         [re quoting initials of poster]
..
RC> >And, warning, some people don't like the way I push
RC> >quotes more quickly off the screen.
RC> 
RC> and some people don't think they're appropriate for Usenet where almost
RC> no clients support them :(  The FIDO readers which support them rely on
RC> seeing leading spaces on the lines involved so as to avoid double
RC> quoting, but this leads to other difficulties when there are spaces at
RC> the start of lines for other reasons.  see news.software.readers passim.

See what I mean, guys?  There's always someone who doesn't
like the way the meal is served.

I didn't invent this idea, I'd seen it a good bit on
posts by other newsreaders in international Usenet groups.
I thought it not only <ahem> cool, but of some utility
when reading deeply nested quotes.  I mailed people to
find out how they did it.  The usual answer was a tricky
editor macro.

However, just because someone on Usenet does something,
that certainly doesn't make it welcome/legal/RFC-compliant/etc.
If you can quote something from a document that strongly
discourages it (e.g. the kind of comments that netiquette
guides make about 4+ line .sigs) I'll chew on it.

Client support?  What's that about?  I just find these
quotes visually helpful.  And what does FIDO have to do
with this?  Perhaps I'm barking up the wrong tree but
I get a feeling, Richard, that you have an emotional
dislike of this style and you're scratching around for
evidence to support your initial gut reaction.

Frankly, when I mentioned "some people" not liking this,
I was being euphemistic.  You're the ONLY person who's
ever told me they don't like it.  And plenty of others
have said they do.

RC> 
RC> >--
RC> >John Washington
RC> 
RC> the space on your "-- " has been eaten by something :(

Yes, the editor I use.  It has many other advantages that
compensate.  Just think of my posts as having no .sig, but
their bodies ending in a couple of dashes and my name.

Do you have a problem with that?

I have considered fixing it by amending Snews to search
backwards from the end of my posts for a "--" within the
last 4 lines, but it's not come high enough up my wish
list yet.  Besides, since I often review outgoing mail
later, some would get eaten again.  I've also considered
a simple utility to scan mail before connecting, but I'm
reluctant to have the additional delay (as well as the
additional effort programming it).  However, if you can
explain to me why it's so important, of course I can
review my priorities.

Why do I get the feeling your post is neither positive nor
constructive?  Too much motorcycle maintenance and not
enough Zen, perhaps?  Drive carefully out there.

--
John Washington

In article <818271966snz@wash.demon.co.uk>, John Washington
<john@wash.demon.co.uk> writes

>If you can quote something from a document that strongly
>discourages it (e.g. the kind of comments that netiquette
>guides make about 4+ line .sigs) I'll chew on it.

"Son-of-RFC1036" 4.3.2 Body Conventions has a SHOULD requirement for
using ">". NB: no other characters, just ">". He notes that ">" should
go onto empty lines as well so that reading agents can easily implement
"skip all quoted text".

>Client support?  What's that about?  I just find these
>quotes visually helpful.  And what does FIDO have to do
>with this?

this style is common (I am told) in FIDO systems. I understand from
other discussions that there is header information which makes it easier
to determine what initials to use, and the editors have suitable code
for handling the nesting.

>Just think of my posts as having no .sig, but
>their bodies ending in a couple of dashes and my name.
>
>Do you have a problem with that?

"Son-of-RFC1036" also has a SHOULD for the .sig separator, and many
reading agents strip them off automajically (which means new users don't
quote sigs - which is a good thing for us all). However, removing a .sig
is hardly a lot of work so its no real problem.

-- 
richard                      Richard Clayton         T U R N P I K E  Ltd
                                                     tel: +44 1306 747747
"Assembly of Japanese bicycle require great peace of mind" quoted in ZAMM

In article <818271966snz@wash.demon.co.uk>
   John Washington <john@wash.demon.co.uk> writes:
> 
> I have considered fixing it by amending Snews to search
> backwards from the end of my posts for a "--" within the
> last 4 lines, but it's not come high enough up my wish
> list yet.  Besides, since I often review outgoing mail
> later, some would get eaten again.  I've also considered
> a simple utility to scan mail before connecting, but I'm
> reluctant to have the additional delay (as well as the
> additional effort programming it).  

I thought about this a while ago. If you were to change your .sig file to 
include "--#" instead of "-- ", it would be fairly easy to write a 
program to do a binary search for the "\n--#\n" string and replace the #
with a space. The advantage of doing it this way is that you can just 
alter the existing file, rather than having to rewrite it, thus making it 
all a lot quicker.

Alternatively, if you always want this signature used, modify your setup 
so that your editor is run via a batch file which adds the signature 
after the edit has completed. 

> However, if you can explain to me why it's so important, of course I can
> review my priorities.

The only obvious "importance" is that it is non-standard, and any 
newsreader which handles .sigs specially will show your .sig to be 
incorrect. As such newsreaders become common, more and more people will 
say "look, that John bloke's got a broken signature". It's a pride thing, 
I suppose :-)

Mark.
-- 
int MeaningOfLife (void) { return 42; }   // mcr@holly.demon.co.uk
function MeaningOfLife :integer; begin MeaningOfLife := 6 * 9 end;

In article <4a59gd$l74@gryphon.looking-glass.org>
   bnb@gryphon.demon.co.uk (Brian Blackmore) writes:
> 
> 1. Its pushes the orginal article out the way quicker.

In theory (ie if everyone used it, which they don't), it would result in 
a three character addition to each line ("MC>") only. When quoted 
material is quoted, it shouldn't be requoted because it already shows who 
the writer was.

Personally, where I have seen this implemented (there are some editors 
which will cope with this sort of thing), several lines alternating 
between one author and another are far less clear this way than they are 
with "standard" UseNet quoting.

That said, I find "> " much easier to read than ">", but again this is 
non-standard. Nobody has complained yet (quick, gimme some wood to touch 
:-)

> 2. It makes writting code to rewrap an article almost impossible and
>    certainly very difficult. Newsreaders are begining to do this now, both
>    the beta version of slrn and bnr (my newsreader) do this.

I have one editor which will cope with pretty much anything you might 
throw at it, and in general it does a very good job.

> 3. It doesn't look right when it itself is quoted.

I don't have a problem with that. I don't like it where it is applied to 
quoted material itself, though:

MR>>Something MR is quoting but didn't write himself....

> 5. I don't like it :-)

This is, of-course, the important thing. The point of it all is to make 
it visually easy to understand for the majority of people. If a 
significant number of people dislike it, then by using it you limit your 
readership. That said, I don't have any problem with John's quoting 
style. 

> 6. IMHO it is visually displeasing.

Isn't this the same point as 5? :-)

The problem tends to come once it gets mangled by someone further down 
the line. In general, everyone having noticably different posting styles 
helps, because as you get to know the people involved the quicker you can 
ascertain whether or not you want to read what is put before you. It 
would be a sad day when all posting stlyes are regimented to the extent 
that there is no freedom for individuals to express themselves. For this 
reason I have nothing against } as a quote mark, although characters such 
as " and . are much harder to read IMHO. Provided it makes what's quoted 
fairly obvious, it's okay by me.

(As for letting intelligent newsreaders work out what is quoted and what 
is not, I would recommend looking at the editor I have seen which can 
cope. IIRC, it goes into some detail explaining the logic it uses to 
decide who wrote what. It is SimTel/msdos/editor/ezquote4.zip (unless a 
newer version is there.))

Mark.
-- 
int MeaningOfLife (void) { return 42; }   // mcr@holly.demon.co.uk
function MeaningOfLife :integer; begin MeaningOfLife := 6 * 9 end;

John Washington <john@wash.demon.co.uk> wrote:
[ deletion ]
>> 2. It makes writting code to rewrap an article almost impossible and

>Rubbish (IMHO).  Whether you writt code or write it, it's
>a SMoP.  You merely need to identify the presence of several
>lines containing something like BB> rather than >.

Have you tried? I have, let me tell you it *is* difficult.

As anyone who has ever tried will tell you, writing any algorthim to wrap
any type of text is in general a pain and always comes up with incredibly
messy and awkward to maintain code. Without ID quoting it is difficult
enough to write a good wrapping algorthim for usenet, in addition to the
normal finding out where the end of paragraphs are, you have identify what
is quote fluff (not easy considering the range of characters that people
use intermixed with spaces, or just spaces, or sometimes spaces), ensure
when the quote fluff changes a new paragraph is forced (does it have a
blank line between the paragraphs like a normal paragraph normally does?),
try to identify lines that have been subject to people rewrapping without
resetting the quoting properly.

And on top of this you want anyone that write word wrapping code for usenet
to have to identify ID quoting (even when in the middle of other quoting
fluff) ensuring that it doesn't catch a stray > which really should be in
the middle of the text. In fact ID quoting is even more of a pain because
if identified incorrectly a real word might get gobbled up as part of the
quote.

Now from my angle a simple > quote sounds really good and people are
perfectly happy with it. It keeps the code simple, it reduces the bugs, it
is more likely to be correct, you have less complaints because this bizzare
article with this bizzare mix of quoting and probably some maths in the
middle didn't quite quote correctly.

[ deletion ]
>> 4. It doesn't match with what son of RFC1036 recommends.. to quote..

>This is the only one of your points that has any validity
>for me.

And you'll see the reason it asks for this is to make automatic processing
of news articles easier.

[ deletion ]
>Summary: much of the argument hinges around emotive issues.
>I don't base my support for this style on visual or aesthetic
>qualities.  In posts with multiple quote levels it helps the
>reader identify which author wrote what.  I think that helps
>avoid confusion (haven't you ever seen flame wars that start
>"I DID NOT WRITE THAT"?).

Summary:

ID quoting is a pain in the arse technically, and has dubious aesthetic
and confusion removal qualities (it doesn't show people quoting people
correctly, and thus removes the age quality of traditional quoting). If
everyone used it then maybe the effort might be worthwhile, but the
likelihood of that happening is slightly less that the sun turning purple
and doing the can-can every monday night. Until this really high probability
event doth occur we will be stuck with ID quoting being traditionally
quoted making it difficult to spot (for both human and machine), and quite
frankly ugly.

-- 
Brian Blackmore                         http://www.wonderland.org/~eternal/

On Thursday, in article <Ek9Y8RAgAsxwEADq@turnpike.com>
     richard@turnpike.com "Richard Clayton" wrote:

> In article <818271966snz@wash.demon.co.uk>, John Washington
> <john@wash.demon.co.uk> writes
>
> >If you can quote something from a document that strongly
> >discourages it (e.g. the kind of comments that netiquette
> >guides make about 4+ line .sigs) I'll chew on it.
>
> "Son-of-RFC1036" 4.3.2 Body Conventions has a SHOULD requirement for
> using ">". NB: no other characters, just ">". He notes that ">" should
> go onto empty lines as well so that reading agents can easily implement
> "skip all quoted text".
>
> >Client support?  What's that about?  I just find these
> >quotes visually helpful.  And what does FIDO have to do
> >with this?
>
> this style is common (I am told) in FIDO systems. I understand from
> other discussions that there is header information which makes it easier
> to determine what initials to use, and the editors have suitable code
> for handling the nesting.

I took your earlier advice to go and take a look at news.(something).

I'm not sure what you were hoping I would find, and I'm not being
deliberately obtuse, but I found the trip rather amusing on two
counts.

1. There are a number of posts in that group using this style of
quoting.  No surprise, really, this crops up all over Usenet
in varying degrees, and tends to be more frequent in the 'techier'
groups, presumably where some users have been able to roll-their-own.

So, the style is common on FIDO, where presumably the technology
is readily available.  It is patchy in Usenet, where the technology
is rare, but when available, appears to be grabbed eagerly enough.
Evidently some (not insignificant) proportion of users welcome
this feature, for reasons we can only guess at.

How often, Richard, have you handed down tablets of stone in this
group, pointing out to people that, just because they wish
something be so, Usenet is not going to change overnight to
accomodate them?  Rather frequently, I believe.

I think your desire for lone '>' quoting, and even Henry Spencer's
not-an-RFC, are just two expressions of personal opinion, with
merit certainly, but as likely to stop the tide as King Canute.

You may convince me (I'm still open to dropping my use of
ID quoting), and you may convince me not to pass on my code.
But how will that serve your cause?  Rob is well capable of
doing this himself.  We're not talking sub-atomic physics here.
And there are many thousands more like him.

2. (and this really had me ROTFL'ing)

While reading the group you sent me to, I stumbled across
a post that implied the Turnpike newsreader has a feature
that highlights quoted parts of articles.  Ho ho ho!  Do
I smell a hidden agenda here?  Do you need some help with
the programming to detect ID style quoting?  :-)

Even if you like GUIs with everything shown in different
colours like a Xmas tree (I don't, but I accept that some
do), this doesn't strike me as a good reason to hold back
a better way of quoting.

Instead, you should be looking at the positive ways forward.
From what you've said about FIDO (of which I'm otherwise
ignorant), there are other things that assist with handling
such quoting.  I don't know what they are, or how they might
translate to Usenet, but my guess would be that something like
an "X-Quoted-IDs: RC>, BB>" in my header would be rather handy
for newsreaders, if it were widely adopted.

Yes, of course this means change (or addition, anyway).
And, of course not everyone who uses unusual quoting
styles will take it up.  However, proposing it now, while
this style is relatively rare in newsreaders, gives it
a much better chance of it being widely implemented,
instead of trying to shut the stable door later.  For,
make no mistake, this style of quote WILL become
commonplace.

You don't believe me?  Imagine Turnpike's marketing
droid comes to you and says "We want this neat new
feature that some of our customers and prospects have
asked for, ID-quoting".  Are you going to turn round
and refuse to put that in?

Maybe you are going to refuse.  I have no doubt of
the strength of your principles and integrity.

And what will then happen when your less principled
competitors start including that feature?

--
John Washington

On 6 Dec, in article <4a59gd$l74@gryphon.looking-glass.org>
     bnb@gryphon.demon.co.uk "Brian Blackmore" wrote:

> [ reguarding funny quote lines ]
>
> There are actually several reasons I could quote for *not* using quote
> lines of the style User>
>
> 1. Its pushes the orginal article out the way quicker.

I mentioned that.  I agree.  TINSTAAFL.

> 2. It makes writting code to rewrap an article almost impossible and

Rubbish (IMHO).  Whether you writt code or write it, it's
a SMoP.  You merely need to identify the presence of several
lines containing something like BB> rather than >.

> 3. It doesn't look right when it itself is quoted.

IYHO.  Not in my HO.

> 4. It doesn't match with what son of RFC1036 recommends.. to quote..

This is the only one of your points that has any validity
for me.

> 5. I don't like it :-)

:-)  I do.

> 6. IMHO it is visually displeasing.

See 5.  You and Richard don't like it.  Rob and I do.
Numerous abstentions of unknown inclination.  Hung jury?
Retrial?

> I should also note that I personally don't know of any newsreaders that can
> actually do this type of quoting, everyone who I've ever seen do it who I've
> asked has done it manually.

So?

Summary: much of the argument hinges around emotive issues.
I don't base my support for this style on visual or aesthetic
qualities.  In posts with multiple quote levels it helps the
reader identify which author wrote what.  I think that helps
avoid confusion (haven't you ever seen flame wars that start
"I DID NOT WRITE THAT"?).

I'm sorry that you find such quoting offensive, and as a
gesture I've used the more traditional style in this post.

--
John Washington

In article <818421965snz@wash.demon.co.uk>
   John Washington <john@wash.demon.co.uk> writes:
> 
> In a perfect world I would have a 3rd editor for Usenet,
> fully key compatible with the two I already use.

Which editor are you using? Surely stripping whitespace is optional and 
can be disabled via some commandline argument? I agree in many cases it 
is useful, but any editor which doesn't have a "do *only* what I tell 
you to do" mode will not last long on my hard disk.

(One of the editors I am currently looking at migrating to will even 
handle binary files intelligently; imagine one of those which stripped 
trailing whitespace :-)

Mark.
-- 
int MeaningOfLife (void) { return 42; }   // mcr@holly.demon.co.uk
function MeaningOfLife :integer; begin MeaningOfLife := 6 * 9 end;

John Washington <john@wash.demon.co.uk> wrote:
>On 8 Dec, in article <4aae85$rjt@gryphon.looking-glass.org>
>     bnb@gryphon.demon.co.uk "Brian Blackmore" wrote:

>BB> John Washington <john@wash.demon.co.uk> wrote:
>BB> [ deletion ]
>BB> >> 2. It makes writting code to rewrap an article almost impossible and
>BB> 
>BB> >Rubbish (IMHO).  Whether you writt code or write it, it's
>BB> >a SMoP.  You merely need to identify the presence of several
>BB> >lines containing something like BB> rather than >.
>BB> 
>BB> Have you tried? I have, let me tell you it *is* difficult.

>That old chestnut, eh?  Negatives are not proofs.

[ deletion ]

>There are two kinds of people in this country, 'can-do's
>and 'no-can-do's, and the latter are taking the
>Great out of Britain.  How about an RFD for demon.ip.status-quo,
>and the rest of us can use a *developers* group for
>discussing development.  Sparking ideas off one another,
>debating the merits of alternatives, & moving forward.

Read any books on development and they will often tell you that one of the
most difficult things to do is to realise when a project won't work and
when it is going wrong. British industry is full of projects that have not
worked or have gone seriously wrong because nobody had the guts to point
out that something is fundamentally fawed so that it can be corrected in
time. The process of somebody telling you that something won't work is
*not* a time to ignore them, it is a time to either prove them wrong or to
find a workaround the problems that they have put forward.

>BB> As anyone who has ever tried will tell you, writing any algorthim to
>BB> wrap any type of text is in general a pain and always comes up with
>BB> incredibly messy and awkward to maintain code. Without ID quoting it
>BB> is difficult enough to write a good wrapping algorthim for usenet, in
>BB> addition to the

>Which is exactly why we need a bit of lateral thinking.

>Assistance in the header would go a long way, I believe.
>I've already mentioned this elsewhere.  No doubt it's
>not the only possible solution.  Let's look for more
>solutions.  Looking for problems, that's easy.  Why
>throw away your life on that?  For Heaven's Sake, get
>out there and try inspiring people.  Get them bashing
>the rocks together, who knows what may come of it.

Correct assistance in the header would go a long way, if it could be made
to stick, my personal opinion is that neither the standard makers nor the
newsreader authors are particulary prepared to support yet another
non-standard header and thus it won't work. Not traditionally quoting an
ID-quote (especially adding spaces) would also go a long way, however I
don't think that every newsreader would end up doing this so it would get
broke eventually.

Now I have taken a long hard look at the problems that ID quoting presents
and IMHO although the problems are not insurmountable, but the risks that
allowing such a method of quoting are not outweighted by the benifits that
the quoting method would allow.

[ Here endeth this conversation, all followups redirected to poster ]

-- 
Brian Blackmore                         http://www.wonderland.org/~eternal/

In article <818560205snx@holly.demon.co.uk>
           mcr@holly.demon.co.uk "Mark Rogers" writes:

>Which editor are you using? Surely stripping whitespace is optional and
>can be disabled via some commandline argument? I agree in many cases it
>is useful, but any editor which doesn't have a "do *only* what I tell
>you to do" mode will not last long on my hard disk.

Any editor which isn't configurable the way *I* want (including
redefinition of any key, including alpha, during editing) won't last
long on my machine.  I haven't found one I like yet - the old IBM PE
comes close, but it strips whitespace from the end of lines with no
'off' for that.  Since that's the action I want 99% of the time, it's
fine for me.

>(One of the editors I am currently looking at migrating to will even
>handle binary files intelligently; imagine one of those which stripped
>trailing whitespace :-)

I don't expect binary editors to do anything...

I did suggest a facility in SNEWS to add the .sig after editing, but it
was thought that it wasn't wanted.  I don't mind (looking through
UseNet, I see more without the --<space> than with anyway, I just don't
put it in at all)...

***********************************************************************
*  chris@keris.demon.co.uk      *                                     *
*  chriscr@cix.compulink.co.uk  *  FIAWOL (Filking Is A Way Of Life)  *
*  100014.3217@compuserve.com   *                                     *
***********************************************************************

In article <4a59gd$l74@gryphon.looking-glass.org>,
Brian Blackmore <bnb@gryphon.demon.co.uk> wrote:

>4. It doesn't match with what son of RFC1036 recommends.. to quote..

7. It's difficult to match with a quote suppression filter on existing
   newsreaders (I use a regular expression to hide all properly quoted
   material and a lot of other such material in trn and only reveal it
   if I need it - that's bad enough because of the number of different
   quote characters used, partly as a result of people trying to defeat
   the following).
8. It defeats the default policy on INN inews to reject articles with more
   quoted than new text.  This is controversial, but, if a news admin 
   conciously selects this option, he wants the default configuration of 
   newsreaders to generate recognizable quoting, in the hope that people
   who have learned how to defeat the filter have also learned to quote
   responsibly.

There is a Unix program, called supercite, which does this sort of quoting.
It's use brought hostile reactions, and seems to have diminished greatly.

On some other points in the thread:

BBS tag lines on USENET postings are not a good advertisement for
yourself, as QWK based readers are associated with broken postings
(which can be partially corrected by intelligent gateways, although
Almac have yet to succeed in this) and because of the low average
quality of such postings in serious newsgroups.

Re the SNEWS deletion question: there are very few things which can't be
done with sufficient effort, so when someone says it can't be done, what
they really mean is that they don't consider the amount of effort to do
it cleanly would justify the return.  However, I do believe that any
attempt to automatically re-wrap articles with arbitrary quoting styles
would always make a significant number of wrong decisions.
-- 
David Woolley, London, England          david@djwhome.demon.co.uk

In article <818421977snz@wash.demon.co.uk>, John Washington
<john@wash.demon.co.uk> writes

[on the JW> quoting scheme]

>How often, Richard, have you handed down tablets of stone in this
>group, pointing out to people that, just because they wish
>something be so, Usenet is not going to change overnight to
>accomodate them?  Rather frequently, I believe.

You have a point here. I think it unlikely that trn and tin will be
using this feature in the near future. Elm does not use it. None of the
Mac newsreaders use it, Agent does not use it, Netscape (some people
seem to believe that to be a newsreader) does not use it. I have no
doubt that some relatively obscure clients use it... you may also have
seen me argue that Usenet is a broad church and it is almost impossible
to make statements about it which are universally true.

>I think your desire for lone '>' quoting, and even Henry Spencer's
>not-an-RFC, are just two expressions of personal opinion, with
>merit certainly, but as likely to stop the tide as King Canute.

I seem to recall that Canute was trying to demonstrate to courtiers that
there were limits to his power. The tale changes in the telling. For
myself, I see an increasing standardisation on platforms ... and that
none of the main players are adopting this style. Spencer is indeed
hoping for change in requiring ">" and not other characters... I think
that he, perhaps alone, may have sufficient power to make an actual
change... but it is perhaps too late for that.

>You may convince me (I'm still open to dropping my use of
>ID quoting), and you may convince me not to pass on my code.
>But how will that serve your cause?  Rob is well capable of
>doing this himself.  We're not talking sub-atomic physics here.
>And there are many thousands more like him.

Useful statements about Usenet are measured in millions - I think you're
beginning to understand my point.

>2. (and this really had me ROTFL'ing)
>
>While reading the group you sent me to, I stumbled across
>a post that implied the Turnpike newsreader has a feature
>that highlights quoted parts of articles.

as indeed does (Free)Agent. Indeed I'm sure a lot of other programs do
too. It's extremely useful and I recommend it to all.

>  Ho ho ho!  Do
>I smell a hidden agenda here?  Do you need some help with
>the programming to detect ID style quoting?  :-)

It's very kind of you to offer, but it seems moderately straightforward
to implement. Automatically deciding which new two/three character pair
to choose, by inspecting the poster's name, is a little more
challenging. BTW I think I'd rather you put in the leading space which
is (in some sense) "standard" in the FIDO world, and avoids various
detection problems.

>And what will then happen when your less principled
>competitors start including that feature?

It will be messed up by the rest of the denizens of Usenet who read news
articles with clients which put ">"s on the front without thinking too
hard as to how appropriate this is. This adds an extra complication to
the next "clever" client which tries to do some unpicking.

richard                                (unaccountably in my front room)
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Though the future's there for anyone to change; you know it seems; it's
easier sometimes to change the past: fountain of sorrow; jackson browne

On Monday, in article <N5rKbsAzGBzwEADd@turnpike.com>
     richard@turnpike.com "Richard Clayton" wrote:

[sig]
RC> no... but if you do have one then it should be separated. This is not a
RC> very good technical definition, but since humans all know a .sig when
RC> they see one it works OK in practice.

The story so far...

1. JW posts some code.
2. A.N.Other asks "can you also post the code for that neat quoting?".
3. RC explains why that style of quoting is undesirable.
4. RC adds an irrelevant swipe at JW's tiny but not-perfectly-formed 'sig'.
5. JW considers the multitude of other sigs around, multiple lines, without
   even a "--", let alone a "-- ", and wonders "Why me???".
6. JW takes the bait and starts a thread to discuss this.

The only constructive thing I'm getting is: in future
I must be more wary of responding to trolls.

Bye.

--
John Washington

On Monday, in article
     <5pck0FAD74ywEA9h@happyday.demon.co.uk>
     richard@happyday.demon.co.uk "Richard Clayton" wrote:

..
RC> to implement. Automatically deciding which new two/three character pair
RC> to choose, by inspecting the poster's name, is a little more
RC> challenging.

Not really.  I don't create them manually.

RC> BTW I think I'd rather you put in the leading space which
RC> is (in some sense) "standard" in the FIDO world, and avoids various
RC> detection problems.

Easily done, but it would then make further quoting go
rightwards faster.  But that space sounds a good idea in
the kind of FIDO-based scheme Chris showed us, which I
found an eye-opener and really attractive.

Anyone got any FIDO source that shows how they do that?
And how well does it cope with a mixture of traditional
Usenet and FIDO styles?

--
John Washington

In article <818773713snz@wash.demon.co.uk>
           john@wash.demon.co.uk "John Washington" writes:

JW>The story so far...
JW>
JW>1. JW posts some code.
JW>2. A.N.Other asks "can you also post the code for that neat quoting?".
JW>3. RC explains why that style of quoting is undesirable.
JW>4. RC adds an irrelevant swipe at JW's tiny but not-perfectly-formed 'sig'.
JW>5. JW considers the multitude of other sigs around, multiple lines, without
JW>   even a "--", let alone a "-- ", and wonders "Why me???".
JW>6. JW takes the bait and starts a thread to discuss this.
JW>

   very accurate.. except of course it really should be, to make sure the
rest of this newsgroup understands what you mean..

1. >>> posts some code.
2. >> asks "can you also post the code for that neat quoting?".
3. > explains why that style of quoting is undesirable.
4. > adds an irrelevant swipe at >>>'s tiny but not-perfectly-formed 'sig'.
5. >>> considers the multitude of other sigs around, multiple lines, without
   even a "--", let alone a "-- ", and wonders "Why me???".
6. >>> takes the bait and starts a thread to discuss this.

     Rob (who's now written his own initial quoting code.)

-- 
      *           <<<:::::::: Rob Coles ::::::::>>>             *
                pgp key via mail pgp@lloyney.demon.co.uk