#1491 ECL lexer regular expression detection code appears incorrect

[Neil Hodgson]

In the ECL lexer, the IsSpaceEquiv function generates warnings from code checkers as the expression contains redundant clauses. The first subclause (state <= SCE_ECL_COMMENTDOC) covers all the later ones.

return (state <= SCE_ECL_COMMENTDOC) ||
    // including SCE_ECL_DEFAULT, SCE_ECL_COMMENT, SCE_ECL_COMMENTLINE
    (state == SCE_ECL_COMMENTLINEDOC) || (state == SCE_ECL_COMMENTDOCKEYWORD) ||
    (state == SCE_ECL_COMMENTDOCKEYWORDERROR)

where

SCE_ECL_COMMENTDOC=23
SCE_ECL_COMMENTLINEDOC=15
SCE_ECL_COMMENTDOCKEYWORD=17
SCE_ECL_COMMENTDOCKEYWORDERROR=18

Unsure of what the correct fix would be.

[Gordon Smith]

Its been a while...

As you probably worked out, this code is based on the cpp lexer (which I am guessing has/had the same warning), in the ECL case SCE_ECL_COMMENTDOC seems to cover all the ECL states (except for a few used in a compare tool).

The safe thing todo would be to change it to: return (state <= SCE_ECL_COMMENTDOC); since that is what it is logically doing anyway, I will investigate what this call is supposed to be achieving.

[Neil Hodgson]

For C++, the first subclause just hits 4 states 0..3: DEFAULT / COMMENT / COMMENTLINE / COMMENTDOC. The styles were renumbered for ECL, so the expression should probably just be a list of the comment styles and the default style. Maybe:

switch (state) {
    case SCE_ECL_DEFAULT:
    case SCE_ECL_COMMENT:
    case SCE_ECL_COMMENTLINE:
    case SCE_ECL_COMMENTLINEDOC:
    case SCE_ECL_COMMENTDOCKEYWORD:
    case SCE_ECL_COMMENTDOCKEYWORDERROR:
    case SCE_ECL_COMMENTDOC:
        return true;
}

[Gordon Smith]

Looking at the code, I think it is only there to handle the special case of an inline regex expression (a JavaScript quirk - is the cpp lexer used for javascript also?).

Based on initial testing your proposed fix would be fine.

[Neil Hodgson]

Yes, the C++ lexer also does JavaScript.

Committed as [17badf].