Re: Scan of regexp mistakes

[Paul Eggert <eggert@cs.ucla.edu>]

On 3/5/19 7:35 AM, Wilson Snyder wrote:
> Paul et al, thanks for this cleanup:
>
>> * commit 3c94c7bc02fcacb4a101188dddb0c44d53444178
>>   Author:     Paul Eggert <eggert@cs.ucla.edu>
>>       Fix regular-expression glitches and typos
> This though failed the verilog-mode.el internal tests due to this:
>
>> -		"\\s-*\\.\\(\\([a-zA-Z0-9`_$+@^.*?|---]\\|[][]\\|\\\\[()|]\\)+\\)\\s-*(\\(.*\\))\\s-*\\(,\\|)\\s-*;\\)")
>> +		"\\s-*\\.\\(\\([-a-zA-Z0-9`_$+@^.*?]\\|[][]\\|\\\\[()|]\\)+\\)\\s-*(\\(.*\\))\\s-*\\(,\\|)\\s-*;\\)")
> Which removed | from the character class.

Actually, that patch didn't remove | from the character class, since
[|---] matches only "-", as the "|--" in the pattern does not match
anything since "|" comes after "-" in ASCII.

However, I see you adjusted the regular expression later to add | to the
character class in a different way, so I assume it's OK now. (I don't
know Verilog.)

It is a little puzzling that the patch caused the verilog-mode.el
internal tests to fail, since the failure couldn't have been due to this
part of the patch. My guess is that some other part of the patch was the
culprit, but whatever it was I guess you fixed it. Sorry about the glitch.

By the way, is it still important for verilog-mode.el to support XEmacs?
I thought that XEmacs was kind of dead. If we didn't need to cater to
XEmacs's regular-expression bugs we could simplify that regular expression.