Re: LAC (lookahead correction) for syntax error handling

[Joel E. Denny]

On Sat, 29 Jan 2011, Joel E. Denny wrote:

> Also, there is another problem with syntax error messages that we've 
> discussed before and that LAC might expose further.  Sometimes, when it 
> seems LAC should correct the expected token list in a syntax error 
> message, both the unexpected token and the expected token list are simply 
> dropped.

Oops.  The unexpected token is still reported.  Only the expected token 
list is dropped.

> The problem here is that, for the sake of translation, the 
> expected token list is limited to 4 tokens.  I'm thinking the user ought 
> to be able to control the limit, possibly removing the limit entirely. 
> Surely we can find some way to handle the translation issues.
> 
> In any case, unless there's a simple fix, I don't expect to address this 
> in 2.5, but I should probably document it.  NEWS should probably at least 
> mention that there are some minor caveats to read about in the manual.