As a part of the stuff I'm doing in bug 100214, the implicit dependency that xpconnect has on the "unicharutil" module is about to become explicit. Basically we need to include nsUnicharUtils.h in order to get the global IsUpperCase that will replace nsCRT::ToUpper on line 2414 of xpcwrappednative.cpp. I say "will" because I'm assuming this bug will be looked at before I finish landing bug 100214... however I have added the inclusion of nsUnicharUtils.h in preparation for this landing.. that's what exposed this dependency.

You need to not break the XPCONNECT_STANDALONE build. This currently means that any dependency on intl stuff needs to be condtionally nop'd if XPCONNECT_STANDALONE is #define'd. This also means making the REQUIRES changes conditional in the makefiles too. I'm wondering if we shouldn't just hack the uses of nsCRT::IsUpper and nsCRT::ToLower in HandlePossibleNameCaseError to just act like the char in question has a high byte of zero and then just use the ascii versions of the methods. I just don't think this debug mode code is going to suffer many (if any, ever! )false positives if we do that.

Let's test the leading char for a high byte of zero, and only fold-and-check if so. No false positives, few false negatives.

Comment on attachment 54588 [details] [diff] [review] I thought of that too. coolness.. I like the idea as well. I'll r=alecf so mike can be the sr=

Comment on attachment 54588 [details] [diff] [review] I thought of that too. Righteous.

oh, and as a side note, I'm getting very close to actually _removing_ XPCOM_STANDALONE :) - just a few more ugly dependencies to clean up.

jband, since you were good enough to produce the patch, you want me to handle the check-in?

dbradley: that would be great. Thanks.

everything has been checked in, marking this puppy fixed.

Marking Verified -