
Sunday 25-Apr-93 04:04:46

--900--

Sure wish we could trap and define [RETURN] in BBSMENU.  We could make
any menu default-on-[RETURN] to any cmd we wanted.

--901--

When BROWSING, it is VERY handy to DROP subs that aren't of interest
to me.  Could [D]rop be made to toggle the setting ON/OFF, for those
"accidental hits".  You could quickly re-join that sub.

> Abort, Drop, Post, Quit here, Skip dir, [CONTINUE]: D
> You are no longer a member of "CNet DOS P-Files"
> Abort, Drop, Post, Quit here, Skip dir, [CONTINUE]: D
> You ARE a member of "CNet DOS P-Files"

--902--

Ken, please consider having IFIDO and XFIDO return values, for better
use in DOS scripts.  Perhaps:

Found new msgs:            return(RETURN_OK);
No new msgs:               return(RETURN_WARN);
Needed dir is missing:     return(RETURN_ERROR);
Control-Panel not running: return(RETURN_FAIL);

We could check for RETURN_OK and QUIT the scripts early.

--903--

CNet> Two levels of NETMAIL access are possible.  The 1st level allows the
CNet> user to USE NETMAIL, provided a destination system is chosen from
CNet> a sysop-supplied LIST of systems.  This list is stored in the
CNet> new file SYSTEXT:SYS.FIDOSITES.  You may edit this, following the
CNet> given format.  The 2nd level lifts this restriction, allowing the
CNet> user to enter ANY netmail address.

I thought level #1 gets its netmail COST from SYSTEXT:SYS.FIDOSITES,
but where does level #2 get its costs?  (Or should they get charged
0?)  Each time I NetMail an address (as a level #1 or #2 user), I get
charged 32000 network credits.  Where do I tweek this?

Also, it would be great if sys.FidoSites supported wildcarding:

>  ;  Valid NetMail distinations:
>  SYSTEM#   ADDRESS     RATE     DESCRIPTION
>
>   1     2:291/715.0    0     Pilot BBS, Belgium
>   2     1:2606/543.0   0     Power Windows, NJ USA
>   3     1:107/590.0    0     AMIGA Lynx, NJ USA
>   4     1:2410/*       200   S.E. Michigan
>   5     2:*/*          500   Europe
>   6     3:*/*          1000  Asia
>   7     *              5000  All others not listed here

I just covered 15000+ nodes, in 7 lines.

--904--

> Mail: Enter a user handle.
> : John Doe @ 2410/555
>
> Invalid FIDO format.  Please use ZONE:NET/NODE.POINT.
>
> Enter the system's address or number (L=List): 2410/555
>
> Mailing John Doe at 1:2410/555.0 (Test Feed)...

The ZONE and POINT should be optional, as stated in the ReadMe.
The 1st attempt at omitting them, fails.
The 2nd attempt at omitting them, succeeds.

--905--

When a user is DLing with the [L]ogoff-after-DL set ON, if he runs out
of byte/file credits, the log-off counter doesn't give him the usual
10 secs to [A]bort or [D]isconnect.  He's just dropped immediately.  I
feel he should get the same prompt/time, as always.  (In fact, he'll
probably need it here, more than ever.)

--906--

There is a discussion Carl Tashian started in the CNET Echo (Ken, I
sure wish you could be a regular part of this), regarding IBM/Major-
BBS.

Their large 100+ user ChatLinks (like CNet's JoinLink) seem to be
occurring regularly.

Should CNet's JoinLink be compatible?  Is MajorBBS willing to allow
other BBS authors to use their method/code of linking?  Could anyone
find out?  Licensing fees?  Would we be limiting CNet's power (current
and future) to IBM/Major's methods?  Or is the design open ended?

--907--

What is everyone using to view a list inside ZOO'ed files.
I've always successfully used "Zoo v2.1b3".

When I [E]xamine things online, CNet executes:
   CNet> zoo >R:petemp0 v "UDBase4:Audio/PlayBeep.ZOO"
and returns nothing.

When I try this by-hand from the CLI:
   CNet> zoo v "UDBase4:12.1/PLAYBEEP.ZOO"
I see the file contents.

When I try this by-hand from the CLI:
   CNet> zoo >R:petemp0 v "UDBase4:12.1/PLAYBEEP.ZOO"
"R:petemp0" gets created, but is "EMPTY".

Seems like ZOO's output can't be redirected, for some odd reason.

I'd like to:
1)  Find out why.
before I
2)  Switch to a different ZOO viewer.

--908--
Ken, I have 6,100 files that I need to place online.
All meticulously AmigaDOS filenote'ed with detailed descriptions.
All sorted into 74 dirs that correspond to those in my 74 file-subs.
What's the quickest way?

I can't AO online because the full-screen editor comes up 6100 times
asking me to edit the filenotes.  (Even if I could overcome that, I
wouldn't want to have to do it 74 times, once per sub.)

Amaint doesn't want to adopt any of them, either.
Am I trying this correctly:
1)  Set each sub's "adopt-at-amaint" flag ON.
2)  Place the files in the correct dirs, on the correct partitions,
    per sub.
3)  Run amaint from the pfile-menu or via events.  (I've tried both.)

Nothing gets adopted anywhere.

Also, anyway for my item's date to pick-up the file's actual datetime
stamp date?   (Some of my 6,100 go back a few years.)

Also, I'd love to use wildcarding when AO asks me for a filename(s).

--909--

S> The point is, any time Ken spends programming a phone book, for example,
S> takes away from time he could be spending improving other parts of the
S> BBS.
S> (Note to Bill... this also applies the a FidoNet front end, which is
S> useless
S> to most people)

Differ people will have different opinions on which things are
"great", "powerful", "useless", or "a waste of time" to add.  (Which
is the way it should be.)

S> a FidoNet front end, which is useless to MOST people..

Not according to a recent (04-Feb-93) VOTE topic.
*MOST* (more than 51%) did NOT vote *USELESS*.
A scant 9.5% said they would stick with TrapDoor.
Those are the facts.  I didn't tell those 100s of people to vote
in the manner that they did.

--910--

"RR" reports:
   CNet> Recovered:    123/   123 items;     324/    324 responses
even if it only recovered data from 30 items and 24 responses,
previously killed.

And sometimes reports:
  CNet> Nothing recovered!
even if it just free'ed-up 100,000 bytes of disk space.

And sometimes reports:
   CNet> Recovered:    123/   123 items;     324/    324 responses
even if nothing was recovered, or killed at all.

Could "RR"'s output be made to read something like:
   CNet> Recovered:  30 of 120 items;   24 of 324 responses;  109032 bytes
or some-such useful #s.

--911--

SpellBinder,

As of Oct-92 you were entertaining the idea of writing your own 1-step,
direct mail tosser/scanner, that would eliminate the current 2-step
TrapToss/xFido/iFido method.  Any progress in the past 6 months?

Even WITHOUT all the bells and whistles, just a simple tosser/scanner
for all those basic Fidonet BBSes out there.

--912--

Brian, am I being overly optimistic hoping that usenet type msgs could be
moved in and out of Cnet's own msg bases just with 2 tiny little 14K
ifido/xfido-like executables?

And eventually change the 2-steps (bundles<->usenet msgs<->cnet msgs) down to
1-step (bundles<->cnet msgs).

--913--


> CNet 2.70:  100 lines possible!  Come on, someone's GOT to try it!

Ken, please consider adding a few #define's to the top of cnet.h.
As CNet expands, so will our tools and pfiles.

#define MAX_LINES       100
#define MAX_GROUPS       24
#define MAX_COMPUTERS    32
#define MAX_USERS     32768
#define MAX_FLAGS        64
#define MAX_TITLE_LEN    32
#define HANDLE_LEN       21
#define REALNAME_LEN     26
#define ADDRESS_LEN      31
#define CITYSTATE_LEN    31
#define ZIPCODE_LEN      11
#define COUNTRY_LEN       8
#define PHONENO_LEN      17
#define PASSWORD_LEN     15

(It would probably make expanding CNet itself, easier, too.)

--914--

Ken, I can quickly test new C-pfiles online with "RUN".  I also wanted
to use it to test a variety of options that my pfile might need.

CNet>RUN PfileTest -h -i -t

They seem to all get stripped off, and the pfile isn't even run at all.
(According to SnoopDos.)
But CNet logs it as having "run".

--915--

It would be great if you COULD allow users to start JoinLink (JL) sessions.
But CNet would need an:
1) "allow-JL-starts" flag for each user.
2) and a "only JL with these systems" phone list file.
(Neither are currently available.)

Actually, all we'd really need is a "JOINLINK <port-#> <phone-#>"
arexx cmd.  The arexx script could do all the security checks, etc.

> You are a level x caller.
> You don't have JL-start access.
> You can only JL from xAM-yPM.
> You can only JL from lines x,y,z.
> You can only JL to these numbers:
> 1)  313-555-2020 BBS Name, Detroit,  MI, 5-lines
> 2)  313-555-1212 BBS Name, Bay City, MI  8-lines
> 3)  609-555-4141 BBS Name, Trenton,  NJ 12-lines
> 4)  717-555-6666 BBS Name, Scranton, PA 20-lines
> 5)  ...
> Pick one:

I'll write the arexx script, if someone can convince Ken to add
a JoinLink arexx cmd.

--916--

Suggestion #916 (and counting) ...

I always forget that I have marked msgs from my previous call(s).

At log-in could we add:
>NOTE: x marked msgs exist. (Use RM to read them.)

onto CNet's current:
>NOTE: 1 file(s) exist in your select buffer. (Use DS to download them.)

--917--

R>  -=> Quoting Bill Allen to Big Brother <=-
R>  > Mail: Enter a user handle.
R>  > : John Doe @ 2410/555
R>  > Invalid FIDO format.  Please use ZONE:NET/NODE.POINT.

R>  Your fisrt attempt had the wrong syntax... no space between the
R>  username and the "@" and the address...

Thanks, that was exactly the problem.

No space between name and "@" is allowed if you use the default zone.
A space is allowed if you specify the zone #.

Perhaps Ken would consider changing that error msg into:
> Invalid FIDO format.  Please use Name@ZONE:NET/NODE.POINT.

Or support both space/no-space formats.

--918--

Actually, I always thought that CNet's current BBSMENU method
reflected a 'hold-over' from CNet's earlier, primitive days.
(Which it outgrew ages ago.)

> BBSMenu group #x, line #y within it, ALWAYS refers to a certain cmd.

That's just begging for trouble:
1)  On heavily aliased/edited/added BBSMENUs.
2)  Each time CNet is updated and cmds are added/moved by Ken.
    (And you have to correctly update your BBSMENU.)
3)  Unreadable things like:  What does *.?,>,<, do at this menu?
    I have to log-in, go try it, just to find out.
4)  Accidently add or remove 1 line and all the following lines are
    'thrown-off' by 1.  (A Big Mess.)
5)  Aliased cmds must come at the top of each menu-cmd line.
    Additional cmds must come at the bottom.
    (Or was it the other way round?)
6)  Menu groups and menu items in BBSMENU can't be arranged for
    better readability.

Each of CNet's built-in call should have a keyword assigned to it.
(CONFIG-ACCTING already has 90% of these keywords named.)
> DOWNLOAD
> EDITMESS
> FEEDBACK
> READMAIL
> SENDMAIL

Each of CNet's menu-types too:
> MAINTMENU
> MAINMENU
> ALLMENU
> SUBMENU

Each line of a readable BBSMENU would have 3 values:
A menu-type word, keyword, and an assigned a keystroke.
> MAINMENU  DOWNLOAD  d
> ALLMENU   MAILVER   mv
> ALLMENU   ACTMONT   am
> MAINTMENU CALLLOG   lc

Any cmd, on any menu, any order, any format, readable in human form, etc.

Solves all of the above problems AND gives us the power to move
things around.

--919--

There will forever be a debate on whether the TERM or the BBS should allow for
different keystroke configuration.

Does Pro Comm allow you to set which keyboard key does backspace?
And what that key does?  And what codes are sent out when it is hit?

If it doesn't, and the BBS you call (Cnet) doesn't either, you're stuck.

If the bbs you call has a "ET" menu like:
> Hit your backspace key now:
> Hit your delete key now:
> Hit your interrupt key now:
> Hit your ESC key now:
and a few others, then you are all set.

Any computer, any term, will then work.

Even if your keyboard:
1)  Has a difficult to reach ESC-key.  (Just reassign/move it.)
2)  Has a broken ESC-key.  (Pick a new 1.)
3)  DOESN'T EVEN HAVE A ESC-KEY ON THE KEYBRD.  (Pick any little-used key.)

S> BA> If the bbs you call has a "ET" menu like:
S> BA> > Hit your backspace key now:
S> BA> > Hit your delete key now:
S> BA> > Hit your interrupt key now:
S> BA> > Hit your ESC key now:
S> BA> and a few others, then you are all set.

S> Sorry, not exactly. It is a great idea, but not really feasable. You see,
S> for something like a visual editor, you must have the ability to
S> specifically place the cursor at a specific cursor location. Sure this

My suggestion will definitely NOT turn non-ansi terminals into full-ansi ones.

Will it help with things like the following?
My backspace-key generates an x value, CNet expects a y.
My delete key is a long, little-finger stretch.  Can't I use this key instead?
Control-C effects the term I am running, can't I use a different key?
My keybrd doesn't even have an ESC key at all!
I need to use ^Q for XON/XOFF, not MCI codes.  (Which I never use anyway.)
I find it easier to remember key-x instead of key-y for that function.
and on and on.

S> BA> > Hit your ESC key now:
S> characters to expect when it says "hit your cursor left key" how would it
S> know how many characters to expect? ANSI cursor left is 3 characters,
S> Commodore cursor left is 1 characters... Who knows what other formats
S> would be like... It just isn't really possible to make a good visual

You know much more about this than I do.
Is it totally impossible to ask the user to hit 1 key and store whatever codes
are sent?  1 2 3 or 10?



S> It is not possible to recognise qualifiers (SHIFT, ALT, etc) with the
S> cursor keys. Actually, it is not possible to detect the state of any
S> qualifier over them modem, period. That is why it has not been done.

>it is not possible to detect the state of any qualifier over them modem
>period.
None?
SHIFT can't be detected?  (How do we get uppercase?)
CONTROL can't be detected? (How does vis-ed work?)

Just so I know, what key combos are we limited to?
shift A-Z
a-z
0-9
puncuation
ctrl a-z
cursor keys

How about:
ctrl 0-9
function keys
keypad keys
ctrl keypad
ctrl function

--920--

> Move more cmds to external executables...

Instead of moving things from internal to external cmds, I'd rather see a
convention that allowed US to alias/move cmds from menu-to-menu within
BBSMENU.  With 400+ lines in it, it'll be nearly impossible to please everyone
with ANY kind of default BBSMENU, that we can't rearrange fully.

Even if Ken made 90% of CNet into a bazillion little executables, there would
always be a handful of cmds that where internal and couldn't be moved around.
(And, of course, those would be the 1s I'd need to move.)

Maybe allow BBSMENU to alias cmds like:
> 2; Available everywhere
> AM~1;AM
> MV~3;MV

Optionally specify the menu-group #.
I just moved a menu#1-type and menu#3-type cmd onto menu#2.

--921--

Does anyone know where the word FLOATS came from?
Anyone seen it used elsewhere?
I always found it to be an odd choice of terms.

Split-vote, Percent-vote, Point-Vote, maybe.

--922--

I had problems with 'AddKeys' when used to 'execute cmds'.  Per Ken,
it doesn't do anything until AFTER your arexx script exits.  Pointless
if you need to:

1)  Do some arexx code.
2)  Run 'AddKeys'.
3)  Do some arexx code.

Ken could forever go nuts trying to add all the new arexx cmds.
(I have a long list myself.)
Maybe some kind of convention (similar to 'getuser' XT#####) could be
used instead.

'RunCmd' 7 3

That would do an internal-call to CNet's BBSMENU group #7, the 3rd
command.

Although, like XT#####, these change and things break.

If we just had some way of:
1)  Tapping directly into CNet.  ("GetUser XT#####" and "RunCmd # #".)
2)  A guarantee that these values would NEVER change.

> ADDKEYS's 'execute only after exit'
> and
> BBSCOMMAND's 'must be global only'
> and
> 'you can only execute those cmds you have access too'
restrictions REALLY tie my arexx-hands in major ways.

--923--

OK, here they are.  A list of things to minimize 'breakage' of pfiles/
arexx/getuser values/etc while CNet grows and expands.  Some are currently
being done, some recently started, others are badly needed.

This is a LONG msg.  You might want to hit SPACE to skip it.

Per CBM, place "$VER:" strings in EVERY executable.  0 exceptions.  If
you forget 1, that will invariable be the 1 important version # we'll
need to know about.  All must be 100% correct also.  It would hurt
to show these in window/screen titlebars, or help screens, etc.

It's well worth remembering to bump the version # for each
new release:
C> static BYTE *version="$VER: v2.60c"

Or if you want something even more automatic, even though this is NOT
the approved CBM format:
C> static BYTE *version="$VER: " __DATE __ " " __TIME__ ;
It would still give us some idea of which newer/older versions
we are using.  (And that line could be installed once in each CNet
executable and NEVER have to be dealt with again.)  c:Version would
report the date/time of compile as "Jan 21 1993 02:21:23".

Full programmer's docs.  I've writing 100s of line of code where I
had to 'guess' at things.  This is a great way to write code that
'breaks' between CNet updates.  (Or even 'breaks' when run on 1
systems VS another.)

A list of guidelines that all programmers MUST follow.
A list of things that Ken guarantee's will NEVER change.
A list of things that Ken guarantee's will only change during
major revisions.  v1.0 v2.0 v3.0 etc

Padding, padding, padding in the structures.  Just a handful of bytes
placed in the middle and end of each structure.

When a MASSIVE # of CNet SysOps make a change to BBSTEXT or BBSMENU,
consider making that change the default to the Master BBSTEXT/
BBSMENUs, too.

MANY more 'GetUser' values.  "Just The Basics" don't cut it anymore.
CNet needs about 200 'fixed' GetUsers values.  XT##### values are
great, but they can change as CNet is updated.  Absolute 'GetUser'
values don't.  Arexx scripts that use these, can work FOREVER.

Document ALL changes made in each release of CNet.  Not 'some' of the
changes.  One unreported change can effect a pfile immensely.

Version #s in many of the structures.  Programmers will be instructed
to check if the port/main/user/item/group/etc structures they are
looking for, are in fact, are the ones they found.  No need to break
all pfiles that work with just the user-data structure, just because
the item-data structure was changed.

Always include the latest VDE.C, CNET.H, CNETFUNCS.H, etc, in EVERY
release.  (I still couldn't get a copy of anything newer than cnet.h
v2.30 as of v2.42e.)

Investigate how other BBS authors are offering things like:
>  Our internal 'C' interface was designed as a
>  standard object-oriented system so the programs you write today, will be
>  100% compatible in the future, no matter how many changes the BBS goes
>  through.
while CNet can not.

Think (or ask users) where to place certain files.  And what to call them.
So they don't need to be renamed and/or moved over and over again.

When it IS necessary to expand the structures and break certain
pfiles, think 1-2 years ahead.  What other features will you be
working on next?  Expand structures 1 'BIG' time per year, instead of
5-10 'SMALL' times.  I'd rather break 100 pfiles, once, and get many
authors to rewrite/recompile, than to break 20 pfiles, and get them to
rewrite/recompile, then another 20, rewrite/recompile, then another 20,
rewrite/recompile, again and again and again.

Expanding the "X" in XT##### GETUSER codes.
1==PortData
2==MainPort
3==UserData
4==ItemType
5==NewSubboardType
6==AccessGroup
7==MessageType
etc, etc
Currently, I believe these can already be obtained but it's a bitch
correctly counting offsets within offsets within offsets etc.  (For
structs that contain other structs.)  And when 1 thing 'moves' (CNet
expansion) nearly everything 'breaks' that appears 'after' it.  With
"X" values for each struct separately, the most that would 'break'
would be that one struct that 'expanded'.

Stick to standards and encourage 3rd party programmers to do so to.
Don't hot-key certain single-key questions, and require others to need
a [RETURN].  Don't permit x-y,z ranges in some places and forbid them
in others.  Just don't.

Document EVERY line of cnet.h.  Even if just 1 phrase per function.

Never use variable names like:  ned, MNewSave, mm. bm, um, s1, s2, s3,
ns, gc, AName, user1, user2, uz, sxc, syc, bz, dz, bn, rn, istart,
iend, br, new_f, new_p, imax, mx, new_ri, gpos, t_new_i, t_new_f,
t_new_p, t_new_r, t_new_n, t_new_ri, rset, nsets, num, num2, tline,
KB1, KB2, iline, nm, rs, new_i, new_n, new_rn sread, mio, listn.  (All
actual names.)  Use short, but appropriate, names which refer to their
usage.

Use SHORT LONG BYTE, never int char and other things that can cause
problems between different C compiler defaults int sizes.

Never impliment something 'good enough for now', that will just need
to be change in the (VERY) near future.  Vote had to have ALL topics,
ALL choices, ALL votes-cast deleted, then again a few months later,
then immediately again shortly afterwards.

Allow programmers to request the files and system-info they need via
very generic C calls to CNet.  (NOT load/save the files directly,
themselves.)  CNet will reply with the correct info regardless of
whether or not it's still 'where it was' 3 (or 40) updates ago.

Ken needs to realize CNet is no longer 'your BBS', 'your baby'.  1000s
of people now depend on it on a daily basis.  You can no longer in
indescriminitly alter things.  Over $2,000,000 of hardware is in use
all over the world, running your code.

With each change you make to CNet, ask how yourself "how many pfiles is
this going to break?".

Ken, I thank you for this (and 1000s of other) opportunities you've
given me to freely speak my mind.

--924--

If the value or cmd in the [] brackets is always the default RETURN choice,
why isn't MAIL's:
>Again, Kill, Quit, Reply, Scan, [Pass]
really:
>Again, Kill, Quit, Reply, Scan, [read current]
or better yet:
>Again, Kill, Quit, Reply, Scan, [read 14]

--925--

Does anyone know the MCI code (if any) to bring up a "More?" prompt
immediately?  I'd like to give users a chance to abort:

> This is a long item.  Do you want to read {MCI}
> More? Y/n/c

--926--

I really have something set wrong, some place.
I played with a variety of ANSI ON/OFF settings, changing my screen
height from 26 all the way upto 44 lines.

In the SCREEN-mode, local, I can't see more 23 lines, before the MORE
prompt appears.

In the WB WINDOW-mode, local, I can't see more than 17 lines at a time.

Is 'screen-ht' being ignored under certain conditions?

--927--

So what would cause an HST-DS 14400 to always connect at "CONNECT 9600/ARQ",
instead of "CONNECT 14400/ARQ" when dialing-out to 4 other 14400 systems?

My modem settings?  Which?
My TrapDoor.CFG settings?  Which?

My current TrapDoor settings are:
> BAUD 19200
> LOCK
> MINBAUD 1200
> NOMAXBAUD
(Although I've tried 100 variations of those.)

Please DON'T post a 500-line CFG file or list of 100 HST-DS parameter settings.
Everything else appears to be working.  I just wanted to tweek the 9600/19200
settings.  Which keywords affect it?

--928--

T> man you folks owe Bill Allen an apology... I have had NO luck with
T> Trapdoor...

No one owes me anything.
Saying a piece of software is 'EASY VS HARD to setup' is a matter of opinion.
Calling someone a cheap/lazy/idiot/liar/jerk in a public forum because you
feel #1 instead of #2, shows everyone a lot about the poster.

T> Trapdoor ow bad I want fido will determin on
T> switching software. I might even put Paragon online instead of
T> trapdoor/welmat as I ran Fido thru it for two years and IT worked..

This is an idea.  Odd but it might work.

I hope Ken sees it as the height of insult for people to resort to using a
buggy/primitive version of ParagonBBS in order to have a frontdoor for a
bugfree/incredible (except no FD) BBS like CNet.

It's a hot, foreign sports car with a lawnmower engine in it.

--929--

S> BB> I am need of a good algorithm for finding "similar" words to any given
S> BB> word. What constitutes SIMILAR?

S> The easiest way to do it is just to find the words alphabetically near the
S> incorrect word. I think this is how Pen Pal did it. It isn't as good as

If the 1st 1-2 chars are wrong, you'll never find a 'good' guess this way.

What you really need here is:
Turn a word's PHONETIC spelling (not the ACTUAL spelling, it's
probably not correct anyway) into a number.  All other dictionary
word's numbers within a range of x values are 'guesses' as to what the
user is really trying to spell.

If "conglomeration" is 38912 and not found in the dictionary.
Show words near that #.

(And I'd suggest adjusting the value of x in real-time.  "Too big" and
users will get dumped on with 200 possible spellings.  "Too small" and
users will get 0 guesses.  It should ALWAYS produce the 5 (or so) closes
guesses.  (Even if the guess are VERY far-fetched, "closest" still
better than "word not found.  No guesses available at all."  That is
a probably with UEDIT's checker, but I'm sure it could be tweeked.)

And definately have a fast, small, pointer index-file, that points
to the main 100,000-200,000 word dictionary.

Ken, 2 possibilities:

1)  An algorithm called SOUND-EX.
    A simple word->number method.
    Many states (IL) have the 1st few #s of a person's driver license
    as the sound-ex code of their last name.

2)  UEDIT's (A GREAT EDITOR!) word-guess spell-checker.
    Executable and C source is now PD.

#1 is a tiny 2k source file.
#2 is a tiny source file, too.  But buried some place in over 1meg of
source modules.  See Fred Fish #800-#840 or so.

I have the source for both.  I'll UL #1 today.  (But my timer is running
out AGAIN!)   Both (or neither) source code might do the job.

Also, did you see my earlier UL of "The 1200 Most Common Words In The English
Language"?  Search that dictionary BEFORE searching the main-dict and you'll
get a VERY large # of hits, quickly.

--930--

Ken, are you also taking suggestions for the look&feel of the spell-
checker itself?

From the editor, hit a speller-key and you are asked to choose:
 1> Real-Time marking.
 2> Real-Time correction.
 3> Current-buffer marking.
 4> Current-buffer correction.

1> Real-Time marking.
   As you type, each time you enter a SPACE, the previous word is
   checked.  If it is NOT found in the dictionary, CNet beeps, and the
   word in question become higlighted.  You can continue to 'type-ahead'
   as fast as you like, as previous words are spell-checked in the
   background.  Errors will be marked as they are found.

2> Real-Time correction.
   As you type, each time you enter a SPACE, the previous word is
   checked.  If it is NOT found in the dictionary, it beeps, and the
   cursor-keys are used to show/pick a new choice.
      A) Up/down.     Move through the 'guess' word-list.
      B) Left/right.  Jump to top/bottom of word-list.
      C) RETURN picks the replacement word.
      D) Any other key exits.

3> Current-buffer marking.
   The current screen is re-drawn.  Each word that is NOT found in the
   dictionary, CNet beeps, and the word-in-question becomes higlighted.

4> Current-buffer correction.
   The current screen is re-drawn.  Each word that is NOT found in the
   dictionary, CNet beeps, and stops at the offending word.  The
   cursor-key searches (see #2 A-D) go into effect.

After each session, if the user chose to keep 'as-is' some words
(even though they weren't in the dictionary), they are asked:
> [D]rop, [L]ist, [E]dit, [S]ave your custom-dictionary words [save]:

Users can also edit their "MAIL:Users/#/_dict" custom-dictionary (as
this is just a straight ascii-file) with a new cmd from "EP".
(It takes precedence over the system's dictionary.)

SysOps can, if they choose, to join all the different "_dict" files
together each x days, and a small arexx script can include/exclude the
words (1-by-1 or all) into the main BBS dictionary.

CNet's speller/dictionary grows smarter and smarter, the more you use
it.

--931--

Every little bit helps.  Mention the Amiga anywhere/everywhere.

All those little cards that fall out of magazines for hardware/software and
even more magazine subscriptions...

Just write across them:
> START SUPPORTING THE 3,000,000+ AMIGA USERS!  1-800-66-AMIGA!

And drop them in the mail, post-paid.

If you sold, (or produced) IBM and Mac hardware and software and you went to
your mailbox tomorrow and saw 53,543 cards that said "START SUPPORTING THE
AMIGA" wouldn't you say to yourself, "Hey, what's this Amiga stuff we are
missing out on?"

But I don't think (nor do I want) the Amiga to ever become as popular as
IBM/MAc.

(SysOps:  Capture the above text.  Put it in your log-out screens.)

--932--

P> BTW, does this mean your finally going to run TrapDoor?? After months
P> of telling you that this would be the best choice if you ever wish to
P> use CNet. i do think you will learn to like TD!! I do.. I have TD run on

I've been 'trying' to run TrapDoor, TrapToss, Foozle, CNet, Ifido,
Xfido, various UUCP doors, file freq tools, et, for months now.  Sure
would be nice to NOT have to pick-and-hunt and piece things together
from shareware authors all over the world just to do networking.  It
can be hard enough WITHOUT adding more difficulty to it.

I was really trying to do this without having to resort to asking questions
about the simplest little things.

>Learn to like it.
I don't think so.  It just does 100x more stuff then I need.
All that gets in the way.

I'm NOT going to buy a keyfile until I can get this running.
Sure wish there was a 'better' way to cripple this thing.
100s of requesters popping up and gadget-clicks just slow me up even more.

I'm afraid I'll just throw in the towel and drop TD long before I even
get it running.  (I've done that 7 times so far.)

I wonder why Ken does put this easy-to-setup frontdoor online here, in
the "few mins" it would take to do.

--933--

OG> 1) Instead of just a HALF screen mode for port screens, how about a
OG> QUARTER screen mode, so you can fit four ports on one screen without
OG> having to go to the ICKY WB mode?

Why is the WB mode so "ICKY"?  I love it.

As for 1/2 1/4 screens...
it would be better to open your own PUBLIC SCREEN (any resolution you like)
and just open as many CNet windows on it, as you need.  Any size, any position
zoom or any size, any position unzoomed.
Even stick the control-panel on that screen also.
Even stick TrapDoors 1-2 windows on it to.
The whole works, on 1 screen, always just 1 front/back gadget-click away.

OG> el1;el2;el3;el4;el5 etc... But RANGES would be a god send, eh?

It has already been suggested MANY times.  (Not that you shouldn't mention it
again.)
"Any place that CNet allows a #.  It also allows a range."
Should be the CNet standard.

As for ANY VDEs, I'd rather just jump up to a sub #, change it, and be allowed
to edit that subs data.  All without even leaving/entering that VDE.

OG> 4) When using the QUOTE a message feature, the QUOTED Part of a message
OG> should be in a different color! Other BBS systems will have the QOUTED

I don't know what's holding up this simple, simple idea.  Except 1 thing.
How would you tell CNet where the quote starts and ends?  The quoted parts of
msgs are just stored as text before/after and in between non-quoted parts.
Maybe have CNet check each line.  Anytime a ">" is found as the 1st 1-3 chars,
then it's a quote.  It would NOT be 100% foolproof, but it might work.  Of
course, the color ON/OFF lines would be a part of BBSTEXT and SysOps could
turn them ON/OFF as they please.

OG> 5) A GREAT Feature in the tradition of CNET would be a WHERE or ACTION are
OG> in the EVENTS section of CNETS config... When Amaint runs, it puts
OG> Maintnence as the WHERE. When a USER does a WHO, they see that ports in

Great idea.  Maybe a new field in BBSMENU for all the non-event run things:
   BBSMENU> FOO | {#3 pfiles:MyPfile}   [WHERESTRING]
WHERESTRING is displayed by WHO anytime a user is running FOO.
(I have a lot more things added onto BBSMENU than onto CONFIG's EVENTS.)

I know pfiles can set the WHERE string themselves, but many programmers
(including me) don't always support it.  And many more (including me) don't
allow the SysOp to set it to anything they like.  [WHERESTRING] would do both,
for ALL pfiles.

OG> 7) Perphas an Option for EP, where users can chose the Default colors for
OG> their calls?

I've been on many CNets where I didn't care for the SysOp's choice of colors.
(I have VERY weird tastes for my WB colors and the colors I use in JrComm to
call BBSes.)  So this idea WOULD be handy.

--934--

D> I've been trying to figure out that situation for a LONG time. Seems Kens
D> given this power to "C" programs.

1) Passed via ONLINE as "CMD THESE THOSE THEM".
2) Passed via ONLINE as "CMD47".
3) Passed via BBSMENU.
4) Passed via text files and {MCI} calls to arexx/dos/C-pfiles.
5) Passed via EVENTS.

I still can't get #1 or #2 working.  (which is what we are talking about here)

Ken needs to impliment ALL ASAP.  And done by 'normal' means that we are all
familiar with:
A) Arexx:       parse arg a b c
B) C pfiles:    argv[2] argv[3] argv[4]...
C) DOS pfiles:  argv[1] argv[2] argv[3]...
D) DOS scripts: .key a,b,c templates        (a la xlha2lha)

And for both:
o) Absolute passing.            "CMD X Y Z"
o) Variable-expansion passing.  "CMD %3 %8 %19"   (Getuser values)

When all 40 work (1-5 in all combos with #A-#D, absolute or
expansion), it'll be 'complete'.

(#1 and #2 are the biggies.  They let you write code just like CNet's
built-in cmds, that users are already familiar with.)

--935--

When you [FO]ward a msg to another user, currently online, I feel he
should get the usual System-OLM-notification.  He currently does not.

> **** System Message
> New mail forwarded from SysOp.
> OLM: Again, Reply, Save, ?=help, [Pass]

--936--

I just set the SORT-ORDER in sub "Bug reports and solutions" to type #2, 3
times. Each time I exited and re-entered the sub, it gets reset back to type
#0.

For several months now I've been trying to find this problem.  It seemed like
I was setting EVERYTHING to type #2 over and over again, but I could never
find a clear-cut example like this before.

On 12-May-93 7pm...

> Updates to 2.0...
 This sub's ORDER just reset itself to "0" AGAIN.
 I've set it to "2" many times.

 Am I the only 1 having this problem?

--937--

1> O!
2> @ Logoff complete
3> {100+ random chars...}
4> NO CARRIER

When I log-off on CNet, turning control back over to TrapDoor, what
would cause #3?  It doesn't seem to be doing any harm, I just haven't
seen that on other TrapDoor/CNets that I call.

--938--

Under CNet v2.63...

A couple of wks ago I accidentally imported the same Fidonet msg packet twice.
That created a handful of items that had duplicate responses tacked onto them.

Even though 14 days have passed, amaint isn't purging these items.

I have that sub's 'amaint-inactive-days' set to 10.
I have that item's 'stop-inactive-purge' set to NO.

Other items without dup-responses ARE successfully being purged in
that sub.

Is there another flag that I have set incorrectly some place?

Is there something significant about these dup-responses that is telling
amaint to specifically NOT purge them?

--939--

Is CNet opening/writing/close "_mmail" over and over for each piece of
mail you choose to save after exiting "MR"?

I wondered what the delay was.

Would it be possible to open/write/close the whole load, just once?

If you have 45 pieces of mail, the delay can be lengthy.

--940--

J> - Implement the use of the 4 level of experience :
J>         - Novice        will get the menu form a systext:menu1 directory
J>         - Intermediate      "        "           systext:menu2     "
J>         - Expert            "        "           systext:menu3     "
J>         - SuperUser         "        "           systext:menu4     "

I love this simple idea!  (I asked Ken about it as far back as 09-Aug-92.)

I've been trying to simulate this with BBSTEXT mods, and Arexx scripts, and
pfiles, and DOS files for ages now.  All have their problems.

It really wouldn't have to be anything harder than "Look for this file, and
display it as the menu, if it exists."

If Ken decides against adding it, does anyone have a GOOD, USABLE work-around?

4 different length menus would do wonders for users not too familiar with CNet
cmds.

--941--

>  for(j=0; j<20; j++)
>     printf("%ld \n", myp->PortZ[line]->SAMNow[j]);

Does that look like the correct way at access the current-user's SAM info?

All values seem to be correct, EXCEPT I get nothing but 0s for
charges, mins-used, mins-idle.

(I am assume the "order" of SAMNow is the same as SAM for the MainPort
and both are in the same "order" as shown by "AM".)

For that matter, my AM output itself, for last-user charges, mins-
used, mins-idle always remains 0, too.

--942--

Ken, any plans on setting up a automated joinlink session for FW, now
that v2.70 has that power?

At 7-8am your graph shows only 18% usage.

Even if just once per week.
Even if just for 1 hr.
It would only take 1 port (that is most likely not in use at the time.)

Just allow incoming JL calls in addition to normal callers.
I'm sure some 10-20 line systems will call FW and link-up.
(Anyone interested in doing so?)

--943--

Uploaded "Guide263.LHA" to FW...

OK, here it is.

Ken's ReadMe converted to a "Manual Supplement" by Jim S.

Now converted into a Point & Click, AmigaGuide, Hypertext document.

Oh, how handy the FULL manual would be in this format.

But I don't know Ken's viewpoint on making disk-based docs, due to
pirating possibilities.

--944--

E> This is kinda weird!  This sub is for "NON-reg C-Net owners?"
E> What kind of C-Net BBS are they running if it isn't registered???????????
E> I didn't think you could run a C-Net BBS if it wasn't registered!

Answering questions and comments from "perspective" (pun intended)
customers can be very important also.

--945--

>We need tick and hatch...

Perhaps several programmers would tackle this task if:
1)  They knew there was a need.
2)  They knew some of the background as to what needs to be done.

A) Do the inbound files just need to be tossed into the correct dirs and let
CNet AO them later?

B) Or should it directly write into the data files thru requests to the
control panel:
data (dir)
  _count                           _Free
  _Headers3                        _Items3
  _Message3                        _short
  _Text
I would *LOVE* to know how to do #B.  For tic-purposes and much, much more.

R>  BA> Perhaps several programmers would tackle this task if:
R>  BA> 1)  They knew there was a need.
R>  BA> 2)  They knew some of the background as to what needs to be done.

R>   Sure theres a need if ya want to import ADS/SAN file echos. (I just
R>   picked them up, so it just became a needed feature for me. :)  With

I don't doubt that there is a need, but like I said, "the PROGRAMMERS need to
KNOW about this need".  (I thought tick/hatch was already available long ago.)

And without full programmer docss to CNet, for AddItem() and AddFile()
callbacks (if those are even available), I don't think we are going to get too
far with this idea.  (How does AO do it?  That is the info that is needed.)

Can someone UL a few tick/hatch files so we can see what needs to be done
here?

--946--

> I'm in need of an arexx script to call-out and exchange files, etc...

Fidonetting allows such automated 2-way msg and file exchanges.

But, if you really want to do-it-yourself via Arexx, I think we will need a
couple of new arexx cmds.  Something that allows:
1)  dialing-out as well as
2)  receiving CONNECT-type strings for interpretation.

Also make a great custom auto-dialer, call-back door, etc.

You might want to take a look at my auto-dialer arexx code online here.
I managed to get #1 working.  And a (nasty) kludge (ugh) that worked around
#2.

Neither one is a good replacement for new arexx cmds that would do #1-#2
automatically.

--947--

Suggestion #947...

I get a lot of mail that I wish to keep for future reference purposes.
But NOT in a bulging mbox.

FORWARDING mail is great.
(But it has to go to a user.  It 'moves' the problem.)

PUTTING or (soon) YANKING mail is great.
(But it is stored in text-format offline, so none of CNet's powerful
online SUB browsing/searching/ranging/etc cmds are available.)

So I really only have 1 choice.
Keep 100s of pieces of old mail in my mailbox.

I'd love to be able to [MO]ve the text "mail msgs" to "sub-posts".

I'd create a few spare "moved msgs" subs:
>  Common Questions and Answers
>  Old Netmail
>  Old Feedbacks
>  Old Posts

> Mbox: Again, Kill, Quit, Reply, Scan, [Pass] : MOVE 4 17
That would MOVE mail #4 into a post in physical-sub #17.
(And delete #4 from my mailbox.)

I get an awful lot of mail and feedbacks that could better be answered
by 100 callers in the public subs, instead of 1 lone SysOp that's
expected to have the time and knowledge to do everything.

And I'm sure a few of the things I do answer in private mails could
be forwarded to public-subs, so I wouldn't have to re-answer them again
next time.

I'd LOVE to see Ken use that feature HEAVILY with all the many CNet
questions he must be answering (over and over again).

I get some pretty important netmail that I really want to save online for
future reference purposes.

> Just save them to a file and AO them...

I just tried this.

From the ReadMail prompt...
> W [RETURN] (write)
> Control-X
> P          (put)
> Base0:DirName/FileName  (How do I avoid unknowningly picking
>                          a filename that already exists?)
> Control-X
> A          (abort)
> K [RETURN] (kill)
> (Wait upto 24hrs until amaint AOs.)

That's a lot of work.  I need to do TONS of msgs.

Even if I did it this way, I'd end up with tons of individual *FILES*
instead of *MSGS* online.  And no subjects, no descriptions, "(NO
DESCRIPTION)" msgs, no "from" username, no way at all to do "Z" or
"ZG" searches, etc.

This idea was meant to make things faster/simplier/easier to do.

I really just wanted to turn email (the current one) into a *msg* post
in a sub item (whatever the next item # in line would be).

--948--

In the spell-checker...
For those of us with ANSI colors turned OFF, would it be possible to
show (instead of trying to invert) the offending spelling error?

> "Malfunktion":  Quit, Change, Guess, Learn, All-skip, [Skip]?

(And to speed up the cursor movement, which wouldn't be needed at all.)

--949--

>Looking to give my callers access to SysInfo output...

Have you seen my "DOSmenu" prg?
It allows a SysOp to specify which DOS cmds users can execute.
Any CLI cmd that outputs text, can be offered to callers.
Speed checkers, CPU checkers, calculators, mem checkers, HD space checkers,
etc.  Whatever you feel a caller is "safe" to run online.

I'll UL the latest release.

--EOF--

Saturday 15-May-93 18:46:09

-Bill "Mr. BBS" Beogelein, 313-473-2020, 2-line HST 14.4k USR DS, 1:2410/207



