[comp.mail.elm] summary of Elm 1.5b bugs

mkhaw@teknowledge-vaxc.ARPA (Mike Khaw) (09/08/87)

I don't recall which of the following I've already reported previously,
so here they all are:


- standout bar has leading and trailing '2' on vt100 - buggy termcap(?)
  or elm's curses fails to interpret curses padding.
- bounce option generates uucp bang sender address even when elm is
  configured for Internet style addresses.
- in "batch" mode, ^C gives a prompt to hit ^C once more (forever!)
  I have an ugly fix for this.
- 4bsd job control:  hangs on a pager pipe (i.e., when pager is not the
  builtin one)
- "from" doesn't understand metacharacters unless blank lines separate
  nonblank lines in .elmrc
- "batch" mode doesn't save copy in sent folder (doesn't read .elmrc)
- needs to use termcap "ti"/"te" if applicable (to force 1 page screen mode
  on terminal (like Concepts) that have more than 1-page screen memory)
- use "ks", "ke" termcap for (vt100) application cursor/keypad keys
- doesn't always obey 4bsd "new" tty conventions (e.g., when in editing a
  "Subject:", ^W should erase the word left of the cursor but doesn't).
  Probably a case of "raw" mode when "cbreak" would've been better.
- needs a "copy" vs. "move" option instead of the current "save" (move) option
- patched version retains read mail in inbox unless deleted or saved elsewhere
- need shorthand for (read) mailbox (i.e, shorthand for ~/mbox)
- options screen input doesn't interpret "~" notation
- mail-only mode needs to have nl mapping turned on for ~h header editing.
- builtin editor miscounts on ^U (kill line), backspacing 1 too many columns
- builtin editor thinks "(" ")" are shell metachars!
- can't modify an alias in the alias database
- from gets sender wrong even when elm gets it right (saved rn news articles)
- changing folders and then changing back to a nonempty inbox kills elm

and a wishlist item:
- automagic ^L screen refresh on return from 4bsd ^Z (I have an ugly way to
  do this using a global flag that the ^Z signal handler looks at).

Mike Khaw
-- 
internet:  mkhaw@teknowledge-vaxc.arpa
usenet:	   {uunet|sun|ucbvax|decwrl|uw-beaver}!mkhaw%teknowledge-vaxc.arpa
USnail:	   Teknowledge Inc, 1850 Embarcadero Rd, POB 10119, Palo Alto, CA 94303

taylor@hplabsz.HPL.HP.COM (Dave Taylor) (09/11/87)

Mike Khaw posts a list of problems with Elm 1.5b...my thoughts on 'em
follow:

>- standout bar has leading and trailing '2' on vt100 - buggy termcap(?)
>  or elm's curses fails to interpret curses padding.

That's fixed.

>- bounce option generates uucp bang sender address even when elm is
>  configured for Internet style addresses.

That's a new one and probably isn't fixed.

>- in "batch" mode, ^C gives a prompt to hit ^C once more (forever!)

That's fixed.

>- 4bsd job control:  hangs on a pager pipe (i.e., when pager is not the
>  builtin one)

That's new, but I believe it's fixed because of some other work I did
with the 4.XBSD signal handling...

>- "from" doesn't understand metacharacters unless blank lines separate
>  nonblank lines in .elmrc

This is a new one...

>- "batch" mode doesn't save copy in sent folder (doesn't read .elmrc)

Deliberate - this isn't a bug - but it can be trivially changed by a
conditional in the routine "save_copy" I believe.

>- needs to use termcap "ti"/"te" if applicable (to force 1 page screen mode
>  on terminal (like Concepts) that have more than 1-page screen memory)

This is arguable though - when I'm on a system with multiple screens of
memory I *don't* want to force it to one-screen mode since then I lose
the ability to move around and possibly use the other screens (for example
to page up and have the message I'm replying to in the upper screen, and
my edit session in the lower screen).

>- use "ks", "ke" termcap for (vt100) application cursor/keypad keys

That's still not in there because of the pain of adding general purpose
conditionals to the already killer case statement in "elm.c".

>- doesn't always obey 4bsd "new" tty conventions (e.g., when in editing a
>  "Subject:", ^W should erase the word left of the cursor but doesn't).
>  Probably a case of "raw" mode when "cbreak" would've been better.

That works now.

>- needs a "copy" vs. "move" option instead of the current "save" (move) option

That is configurable -- if you're not in "berkeley mail emulation mode"
(e.g. "resolve=ON" in your .elmrc) then S)aving a message will not result 
in it being marked for deletion.

>- patched version retains read mail in inbox unless deleted or saved elsewhere

That's a local problem - I've never had this problem myself.  Wait until
1.7 is released, please.

>- need shorthand for (read) mailbox (i.e, shorthand for ~/mbox)

That's a good idea but isn't there.  Though it DOES have "%" and "!" as
shorthand for your incoming mailbox (e.g. C)hange mailbox "!" will get
you into your default incoming mailbox).

>- options screen input doesn't interpret "~" notation

True enough.

>- mail-only mode needs to have nl mapping turned on for ~h header editing.

That's fixed.

>- builtin editor miscounts on ^U (kill line), backspacing 1 too many columns

That's fixed.

>- builtin editor thinks "(" ")" are shell metachars!

I don't understand what you're talking about here...is this with regard
to how it does ^W backspacing?  Or...?

>- can't modify an alias in the alias database

That's really a whole can of worms you're opening here, Mike!  I would
like to see Elm have a general purpose database facility for the aliases,
including searching, updating, adding, deleting, and so on, but it just
isn't there...

>- from gets sender wrong even when elm gets it right (saved rn news articles)

Again, this is news.

>- changing folders and then changing back to a nonempty inbox kills elm

Not on my machine.  Perhaps this is an artifact of your patches?

>and a wishlist item:
>- automagic ^L screen refresh on return from 4bsd ^Z (I have an ugly way to
>  do this using a global flag that the ^Z signal handler looks at).

This is a bit friendlier on the new version, but since I'm not using curses
for Elm (that's be a NICE thing to add) it's rather difficult to know at any
given time A) where we are and B) what the screen looked like at that point.
The program also hops out of screen mode into line mode and back, so that
would just compicate things too (e.g. the builtin editor from the main screen).

In summary, most of what Mike is talking about is fixed in the latest
version, and those few things that aren't would either be fairly easy
patches for someone to make or are sort of slung along as part of the
fundamental paradigm.
					-- Dave Taylor