source location should be in characters not bytes

[Derick Eddington]

... because of multi-byte encodings. The below error tells me byte 230 but I really want to know what character. When I tell my editor to go to character 230, it goes to "d" in the comment.

[d@eep:~/t6]-> cat source-location.ss
(import (rename (rnrs) (lambda λ)))
(define a (λ () 1))
(define b (λ () 1))
(define c (λ () 1))
(define d (λ () 1))
(define e (λ () 1))
(define f (λ () 1))
(define g (λ () 1))
(define h (λ () 1))
(define i (λ () 1))
(f oops) ;abcdefghijklmnopqrstuvwxyz

[d@eep:~/t6]-> ikarus --r6rs-script source-location.ss
Unhandled exception:
 Condition components:
   1. &who: oops
   2. &message: "unbound identifier"
   3. &undefined
   4. &source-information:
       file-name: "source-location-bug.ss"
       character: 230
   5. &trace: #<syntax oops [byte 230 of source-location-bug.ss]>

[Abdulaziz Ghuloum (aghuloum)]

Use vi :-)

Seriously, in vim, ":goto 230" goes to oops. So, you either find a way to configure your editor, or I have to find a way to configure vi *and* fix this bug. What do you use anyways? Your code doesn't look like it was inflicted by emacs.

[Michael D. Adams (mdmkolbe)]

Or the error message could report both the character and byte position so users can just use the right one depending on whether they are using a broken and or a non-broken editor.

[Derick Eddington (derick-eddington)]

[Whoa, that's weird; I emailed a response to Aziz's comment 2 days ago but it seems LaunchPad lost it.]

-------- Forwarded Message --------
From: Derick Eddington <email address hidden>
To: Bug 226090 <email address hidden>
Subject: Re: [Bug 226090] Re: source location should be in characters not bytes
Date: Sat, 03 May 2008 04:56:57 -0700

On Sat, 2008-05-03 at 10:57 +0000, Abdulaziz Ghuloum wrote:
> Use vi :-)

I've tried, a number of times... :) Once the acceptable-looking
anti-aliased GTK2 version came along I really wanted to, but I couldn't
get over how much I hate its scripting language and how it doesn't have
the best S-expression intelligence (at least what I could find out
about). I also tried Cream but had the same two issues.

> Seriously, in vim, ":goto 230" goes to oops. So, you either find a way
> to configure your editor, or I have to find a way to configure vi *and*
> fix this bug.

But don't you think it makes more sense for it to be in terms of
Scheme/Unicode characters, since R6RS source code is specified to be
Unicode?

> What do you use anyways? Your code doesn't look like it
> was inflicted by emacs.

Hahaha. DrScheme. As it's nearing complete R6RS support +
*.IMPL.{ss,sls}, I can run my code in it, and use Dr's syntax checker,
debugger, macro stepper, and help system (which uses the R6RS docs),
and it's got pretty nice S-expression intelligence and syntax coloring
and formatting :)

If you post to ikarus-users about your Scheme-ing vim setup and how you
use it and maybe some of the custom debugging tools you might have, I'll
consider trying vim again :) Else, if this stays as bytes, I guess I'll ask the
PLT guys how to make a byte locator.

I'm not really concerned about this right now, but given we want the
whole world to use Ikarus and they'll want to use multi-byte characters
a lot more than me...

[Derick Eddington (derick-eddington)]

I asked the vim-multibyte mailing list how to go to character positions, and so far I've gotten this response:

==========================================
On 03/05/08 23:14, Derick Eddington wrote:

> I'm trying to figure out how to configure Vim so that ":goto X" will go
> to character X not byte X. If I have an external thing (in my case, an
> interpreter) telling me I need to look at character X, when the file is
> in a multi-byte encoding (typically UTF-8), I need to be able to tell
> Vim to go to character X. How can this be done?

> I searched around for the answer but couldn't find it, so thanks for any
> help.

":goto" will always count by bytes, but there's another way to do it:

        :set ww+=s ve=onemore
        :exe "normal gg" . (char_number - 1) . " "

don't forget the space at the end. When doing it at the keyboard you may
type just gg then the number (one less than the desired character
number) followed by a space: "exe 'normal'" is not necessary in that case.

see
        :help 'whichwrap'
        :help 'virtualedit

The ":set virtualedit=onemore" is necessary to count linebreaks (but
each of them as only one character even on Windows). That option value
doesn't exist in Vim 6 or earlier.

You can check the result with the Normal-mode command g Ctrl-G
==========================================

I've since asked about how to turn this into a Vim function/command available upon start-up. If you don't already know how, I hope to get a response, see the thread:

http://groups.google.com/group/vim_multibyte/browse_thread/thread/6a2b7a871faf851d

[Abdulaziz Ghuloum (aghuloum)]

Fixed in revision 1471.

[Abdulaziz Ghuloum (aghuloum)]

On May 5, 2008, at 12:48 PM, Derick Eddington wrote:
> [Whoa, that's weird; I emailed a response to Aziz's comment 2 days ago
> but it seems LaunchPad lost it.]

Launchpad has been doing that a lot lately.

> But don't you think it makes more sense for it to be in terms of
> Scheme/Unicode characters, since R6RS source code is specified to be
> Unicode?

Maybe. We'll get into more issues in the future as we go along since
there are some more transcoding/normalization/eol-style thorns just
around the corner. A byte was just simpler since we all agree on
what bytes are, but oh well, too late now. :-)

>> What do you use anyways? Your code doesn't look like it
>> was inflicted by emacs.
>
> Hahaha. DrScheme. As it's nearing complete R6RS support +
> *.IMPL.{ss,sls}, I can run my code in it, and use Dr's syntax checker,
> debugger, macro stepper, and help system (which uses the R6RS docs),
> and it's got pretty nice S-expression intelligence and syntax coloring
> and formatting :)

That's good if you've found it useful. I may be too old to use
something that was not first invented around the time I was born
(scheme:1975, vi=1976).

> If you post to ikarus-users about your Scheme-ing vim setup and how
> you
> use it and maybe some of the custom debugging tools you might have,
> I'll
> consider trying vim again :) Else, if this stays as bytes, I guess
> I'll ask the
> PLT guys how to make a byte locator.

Neah, I'll skip the tool advocacy/zealotry for now.

Aziz,,,

[leppie (leppie)]

> That's good if you've found it useful. I may be too old to use
> something that was not first invented around the time I was born
> (scheme:1975, vi=1976).

I was born in '75 and I do not have issues with newer tools :) Then again, I have only been programming seriously for 6-7 years.