Subject:
|
Re: Backwards Compatibility (Was Calling all Meta-commands)
|
Newsgroups:
|
lugnet.cad.dev
|
Date:
|
Sun, 16 Mar 2003 18:46:25 GMT
|
Viewed:
|
5442 times
|
| |
|
|
In lugnet.cad.dev, Dan Boger writes:
> On Sun, Mar 16, 2003 at 05:54:10PM +0000, Kevin Clague wrote:
> > I think we cannot ignore the backwards compatibility issue though.
>
> right.
>
> > What we want is an explicit way to differentiate comments from
> > meta-commands. I think defining an explicit mechanism for comments is
> > completely backward compatible, because if you do not recognize the first
> > token in a type 0 record, it is a comment.
> >
> > By using (META) or {META}, you've caused more hassles than you've solved.
> > Now we programmers have to change our programs to parse type 0 records in
> > two ways: The way we have today ("0 token"), *and* the new way ("0 (META)
> > token").
>
> but you just said you're doing that anyway - "if you do not recognize
> the first token in a line type 0 record, it is a comment". Is adding
> '{META}' to the list of recognizable tokens an issue? Also, you don't
> have to add it - if you don't, the line is ignored, just like any other
> comment. Perfect!
The exact same argment can be used in favor of dillineated comments. ;-)
>
> > It makes a lot more sense to me to provide an explicit way to
> > differentiate comments using some keyword (COMMENT, //, #, !, etc.) You
> > document the magic comment token so that no one can ever define that as a
> > meta-command, and you greatly reduce the probability that someone's first
> > token in a comment line is mistaken as a meta-command.
>
> wouldn't that break backwards compatibility with ldraw.exe? Or does it
> recognize any of those as a valid, but ignorable line?
Nope. Any program that does not recognize // (picking one as an example)
simply ignores it. Just like (META).
>
> > It is very common in programming language to differentiate comments from
> > language significant grammar using comment start tokens than it is to
> > identify the things the computer is supposed to pay attention to (i.e. (META)).
>
> right, but it's not unhead of to use special comments as "hints" to the
> program - like "#define", etc.
I don't want to get sidetracked here, but #define is not a special comment.
Our current case with type 0 records is exactly the case you outlined above.
Meta-commands are special comments used as "hints".
The only real problem that arises is when the user starts a type 0 record
with a token that happens to be interpreted as a meta-command. They might
get an error message that completely confuses them. Comments can cause
errors ?!?!
If we treat // as a meta-command that means comment, and people use them,
then we eliminate the issue without changing a single line of code.
>
> > Using (META) causes more work for us programmers and really doesn't solve
> > much of anything.
>
> That might be true - I'm not one of the programmers who has to implement
> these, so I don't get a say here. I do think it solves the problem of
> accidental meta-commands in unintentional comments.
As does having people switch to // to specifically delineate comments.
>
> :)
We can even continue with the status quo, but there will always be confusion
about what is a comment and what is a meta-command. If we go for // as a
comment indicator, we will always have two forms of comments: implicit
comments, and explicit comments. If we got for (META) as a meta-command
indicator, we will always have two forms of meta-commands: implicit and
explicit. We always run the risk of these two problems:
1) Someone creates a new meta-command that falsly triggers on someones old
comments. (avoidable by making all new meta-commands start with (META))
2) Someone creates a comment that starts with a token that someone had
already used as a meta-command. (avoidable by making new comments using //).
Thus the nature of backwards compatility.
Maybe we need to do both?
I'd personally prefer something like ! instead of (META) or {META}, but that
is because I'm used to the concept in unix.
Kevin
|
|
Message has 1 Reply:
Message is in Reply To:
154 Messages in This Thread:
(Inline display suppressed due to large size. Click Dots below to view.)
- Entire Thread on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
This Message and its Replies on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
|
|
|
|
