| abuild 1.1.3 | <– Date –> <– Thread –> |
|
From: Jay Berkenbilt (ejb |
|
| Date: Sat, 2 Oct 2010 09:57:30 -0700 (PDT) | |
I have released abuild 1.1.3 on sourceforge.net and have updated the documentation on abuild.org. This version is fully compatible with the other 1.1.x versions. Below are the release notes for version 1.1.3. Please see the abuild manual for additional details. You can find the documentation links at http://www.abuild.org/. --------------------------------------------------------------------------- â Output Capture â Abuild is now able to capture the output of builds and associate each line of output with the build item that produced it. For additional details, please see ChapterÂ20, Controlling and Processing Abuild's Output. â It is now possible to have abuild prefix every line of normal output and/or every line of error output with fixed prefixes. For details, see ChapterÂ20, Controlling and Processing Abuild's Output. â Bug Fixes â File-specific variables for XCPPFLAGS, XCFLAGS, and XCXXFLAGS were referenced in a manner that prevented them from being properly expanded. They are now properly expanded, so their values may include references to other variables. â Miscellaneous Enhancements â The codegen-wrapper command now accepts the --normalize-line-endings flag, which tells it to disregard differences in line endings when checking cached files to see whether their sources have changed. Thanks to Jeremy Trimble for the suggestion. â When a platform plugin's list_platforms script had Windows-style line endings, abuild (or, more accurately, the underlying system) would produce a confusing error message when trying to execute the script. On non-Windows systems, abuild now explicitly calls your attention to the incorrect line endings if list_platforms fails. â The first line of output produced by the processing of any build item now always includes an indication of the build item name and output directory. In prior versions, there were certain rare instances in which this would not happen. For example, if an interface-only build item depended on two other items whose Abuild.interface files declared conflicting variables, abuild would complain about the conflict and indicate where it occurred, but it would not provide any hint as to what build item caused the two interface files to be loaded together. Now abuild will always indicate which build item is responsible for causing the problem to be detected. â When a platform selector specifies a platform type, platform, compiler, or option specification that doesn't match any items anywhere in the entire forest, abuild now reports that as an error. It remains (and must remain) perfectly normal for platform selectors to apply to only a subset of the trees or items in a forest since most plugins only apply to subsets of the forest. The previous behavior of ignoring invalid platform types in platform selectors was intended to allow the same platform selectors to work across multiple forests, but in practice, having abuild tell you about potential typos in platform selectors is much more important functionality, and it's not really practical to use the same platform selectors across multiple forests in general anyway. â Documentation Changes â New help topics, make and groovy, provide brief reminders of things you can do in Abuild.mk and Abuild.groovy files or local rules files that they reference. â Fix errors in documentation for global plugins and platform selectors. â Create new section on capturing and parsing abuild's output. --------------------------------------------------------------------------- -- Jay Berkenbilt <ejb [at] abuild.org>
- (no other messages in thread)
Results generated by Tiger Technologies Web hosting using MHonArc.