RESOLVED DUPLICATE of bug 10553 10343
Windows portability fixes
https://bugs.webkit.org/show_bug.cgi?id=10343
Summary Windows portability fixes
Krzysztof Kowalczyk
Reported 2006-08-11 00:16:50 PDT
1. CSSValueKeywords.in has empty lines which, they end up in CSSValueKeywords.gperf because makevalues script doesn't filter them out, this makes gperf fail and the build will fail because CSSValueKeywords.* are not properly generated Fix is to remove empty lines (as in attached patch) or fix makevalues script to filter them out. 2. ENCODING_PREFIX in build-generated-files.sh needs to be defined as "\" \"", otherwise it'll end up being nothing in DerivedSources.make and the followig line will fail: CharsetData.cpp : platform/make-charset-table.pl platform/character-sets.txt $(ENCODINGS_FILE) perl $^ $(ENCODINGS_PREFIX) > $@ A comment about that is in make-generated-sources.sh (but it's probably not used anymore). In general, this is very fragile so even a better fix would be good (one that adds some diagnostics to scripts - diagnosing build problems due to one of those scripts that auto-generate files failing isn't easy). Also, it would be nice if VS build would stop if pre-build event fails, because going on is usually pointless since it'll stop at some much later point with obscure error. Unfortuantely, I don't know the required VS magic (if one exists).
Attachments
Fixes for the 2 problems (2.33 KB, patch)
2006-08-11 00:17 PDT, Krzysztof Kowalczyk
darin: review-
Krzysztof Kowalczyk
Comment 1 2006-08-11 00:17:43 PDT
Created attachment 9975 [details] Fixes for the 2 problems
Eric Seidel (no email)
Comment 2 2006-08-11 13:58:53 PDT
Comment on attachment 9975 [details] Fixes for the 2 problems The second fix looks fine. The first fix really should be to fix the script and not make the data file harder to read.
Darin Adler
Comment 3 2006-08-15 21:17:28 PDT
Comment on attachment 9975 [details] Fixes for the 2 problems I agree with Eric's comment. The fix for build-generated-files.sh is fine, but the fix for CSSValueKeywords.in should be in the script, not the source file. Since this one patch contains both, review- for now.
Krzysztof Kowalczyk
Comment 4 2006-08-24 20:29:56 PDT
Patch in 10553 has this fix too. Trying to consolidate the changes into one bug. *** This bug has been marked as a duplicate of 10553 ***
Note You need to log in before you can comment on or make changes to this bug.