User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.9a1) Gecko/20051219 Firefox/1.6a1 Build Identifier: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.9a1) Gecko/20051219 Firefox/1.6a1 During the testing of bug 318922 many inconsistencies where found in the use of {} expression value substitution in E4X syntax in the Mozilla implementation. Example // test 1 b='b'; a=<a xmlns:p='http://a.uri/'> <p:b{b}>x</p:bb> </a>; is valid per mozilla implementation But // test 2 b='b'; a=<a xmlns:p='http://a.uri/'> <p:{b}b>x</p:bb> </a>; gives SyntaxError: invalid XML qualified name Reproducible: Always Steps to Reproduce: go thru attachment e4x_inconsistancy.html Actual Results: some INVALID SYNTAX other VALID SYNTAX Expected Results: Expect for the last test, all of the should be either INVALID SYNTAX or all should be VALID SYNTAX to avoid any inconsistency.

Thanks for this report -- it's an easily fixed bug. /be

Note use of comma operator to unobfuscate the dependent nested boolean expression. If (xml-only-mode or next != '{') then insist on nextc being a valid NC. /be

Comment on attachment 206872 [details] [diff] [review] proposed fix Sure.

Fixed on trunk. Waiting for shaver's sr= and some baking before nominating for 1.8.0 branch. /be

great.. I also see another inconsistency in E4X see bug 321564#c3

Checking in 11.4.1-07.js; /cvsroot/mozilla/js/tests/e4x/Expressions/11.4.1-07.js,v <-- 11.4.1-07.js initial revision: 1.1

tested on Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.9a1) Gecko/20051227 Firefox/1.6a1 Looks Better.... Issues // test 3 u='http://a.uri/'; p='p'; a=<a xmlns:pp={u}> <{p}p:b>x</pp:b> </a>; Valid !!! // test 4 u='http://a.uri/'; p='p'; a=<a xmlns:pp={u}> <p{p}:b>x</pp:b> </a>; SyntaxError: illegal XML character // test 5 u='http://a.uri/'; p='pp'; a=<a xmlns:pp={u}> <{p}:b>x</pp:b> </a>; SyntaxError: illegal XML character 6 : valid 7, 8: invalid !!! 13, 15, 16, 17, 18, 19 : invalid (let me know whether those test are also valid)

More work needed -- thanks for keeping after this! /be

(In reply to comment #9) > More work needed -- thanks for keeping after this! No probs. As far as I concern {} sutff is working as required for all most all possible common use. And any software always will have the last bug. One question: Will javascript will be one day available as an OS shell script language like AWK, perl, tcl, Python? Hey there is also another good use for {} syntax javascript. Using it one can create a Mail Merge like stuff. NAME = "Biju"; STREET = "123 Blue heaven"; CITY = "The Atlantis"; STATE = "ZA"; ZIP = "97887"; a = <>Dear {NAME}: Our records show that your address is: {STREET} {CITY}, {STATE} {ZIP} If this is incorrect, please let us know immediate so that we can update our database. And send all junk mails to you every week. Thanks, Bob www.TheJunkMailSender.com </>; a;

Comment on attachment 206872 [details] [diff] [review] proposed fix sr=shaver, now that I've unwound the state space in my head a little. Ah, good to be back.

Would consider approving a baked patch.

Comment on attachment 206872 [details] [diff] [review] proposed fix This has been baking on the trunk since 12/26. /be

Patch makes things better only, not worse (for workarounds, e.g.). /be

Comment on attachment 206872 [details] [diff] [review] proposed fix a=dveditz for drivers, land quickly

Comment on attachment 206872 [details] [diff] [review] proposed fix I landed this patch on the 1.8* branches. /be

Really should spin out the remaining problems into a new bug report at this point, since we've landed the patch. Biju, can you do that? Thanks, /be

fixed per comment 5. Fixed on branches per comment 16

v.fixed on 1.8.0.1 with Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.8.0.1) Gecko/20060111 Firefox/1.5.0.1 using testcases in comment #1, results match those found to be acceptable in comment #8.

v 2006-01-11 1.8.0.1, 1.8.1, trunk windows/linux/mac

(In reply to comment #17) > Really should spin out the remaining problems into a new bug report at this > point, since we've landed the patch. Biju, can you do that? Thanks, We really need the other testcases unfixed by this bug's patch spun out into another bug. Bob, could you please do that? Thanks, /be

(In reply to comment #21) will do.

filed bug 325750