Review Request: KJS: No longer treat invalid hex as string

[Bernd Buschinski]

This is an automatically generated e-mail. To reply, visit: http://git.reviewboard.kde.org/r/104076/

Review request for kdelibs. By Bernd Buschinski. Description KJS: No longer treat invalid hex as string Fixes ecmascript - S7.8.4_A6.1_T4 (invalid hex \x0G) - S7.8.4_A6.2_T1 & S7.8.4_A6.2_T2 (too short hex \xA) Testing tested with ecmascript, no new regressions Diffs kjs/lexer.cpp (e89de5f) View Diff

[Maks Orlovich]

This is an automatically generated e-mail. To reply, visit: http://git.reviewboard.kde.org/r/104076/

This is consistent with the spec; but I am worried that in the wild behavior may not match the spec yet (at least it doesn't for Chrome 18; it may be safe since FF seems to follow this, but I don't have a good version at spec now)

kjs/lexer.cpp (Diff revision 1)
int Lexer::lex()
320	setDone(String);	320	setDone(String);

Any reason you're still permitting this recovery (testcase: "\x")?

- Maks

On February 25th, 2012, 4:31 p.m., Bernd Buschinski wrote:

Review request for kdelibs. By Bernd Buschinski.

Updated Feb. 25, 2012, 4:31 p.m.

[Bernd Buschinski]

This is an automatically generated e-mail. To reply, visit: http://git.reviewboard.kde.org/r/104076/

On March 20th, 2012, 9:18 p.m., Maks Orlovich wrote:

kjs/lexer.cpp (Diff revision 1)
int Lexer::lex()
320	setDone(String);	320	setDone(String);

Any reason you're still permitting this recovery (testcase: "\x")?

Because I feared it would break even more, or rather there was some string appending magic that would make it valid.
In but there should be no way, \x must be followed by 2 valid hex values. I will update the patch

- Bernd