#66 Wtkprerocess changed parsing behavior with //# prefix

[Anonymous]

Wtkprerocess changed parsing behavior, breaking backward-compability

Antenna 0.9.5 uses '//#' token to distinguish preprocessor directives.
Instead of this, in 1.0.0 preprocessor allows whitespace between comment '//' and '#' sign, thus breaking compilation of such code:

1) commented out C-style preprocessor
// #if !defined(FLAG)
2) normal comments starting with '#'
// #r for CR, #n for LF

=====================================================
From CVS:

Preprocessor v1
antenna/src/antenna/preprocessor/v1/PreprocessorLine.java
---------------------------------------------------------
157 space = s.substring(0, p);
158 s = s.substring(p);
159
160 if (s.startsWith("//#")) {
161 type = parseCommand(s);
162 }
163 else {
...

Preprocessor v2
antenna.preprocessor/src/antenna/preprocessor/v2/parser/APP.g
-------------------------------------------------------------
52 PREFIX : "//" (WS)* "#";

antenna.preprocessor/src/antenna/preprocessor/v2/parser/HELPER.g
----------------------------------------------------------------
36 PREFIX : "//" (WS)* "#";

[Omry Yadan]

Logged In: YES
user_id=126268
Originator: NO

This is by design:

when you format your code with the eclipse formatter with the default
settings, it puts a whitespace between the // and #.

for example, this:

//#if !FOO
int x = 0;
//#endif

would turn into this after formatting the code:
// #if !FOO
int x = 0;

// #endif

I can only suggest that you use /* */ comments for that very special
case where your comment begins with #, or just use a different first
character. for example, instead of:

// #r for CR, #n for LF

you can have:

// Use #r for CR, #n for LF