fix: filterwhatcomp and filtercheckcomp - correctly strip absolute directory and replace with EPOCROOT on windows.

fix: failure of make engine test when a make engine alias contains a modifier. There can be more than one variant per alias via "." e.g. make.extrastuff where make and extrastuff are variants.

Fix up file testing with checkwhatcomp and make epocroot driveletter stripping case insensitive.

FilterCheckComp - checking file existence by component

Merge

release: notes for sbs_filter-related changes

Tests cor check and whatTests cor check and what

fix: make filters work with new default python fix: Make Check filter actually check stuff when run from the filter

Retain MakeEngine class API. Why not?

review comments, dates, test ids

Merge

test for make engine validation

don't break sbs_init.xml files with make engine validation changes

fix: get make engine name validation working with aliases.

fix: better error messages when an incorrect make engine is specified. Requires that all make engine variants should extend "make_engine".

Oops - creating new release notes section for that new fix

Catch up

Release note: Bug 1569 - excessive recompilation in incremental tracecompiler builds

Bug 1569 - Making test case skip on Linux while TC doesn't support Linux

Bug 1569 - review improvements - making tc_incremental more robust by cleaning first and faster by building urel instead of udeb

Fixing Bug 1569 - excessive recompilation in incremental tracecompiler builds

Adding extra source file to TC winscw test code, and fixing it so it works

First drop of untested winscw TC code. Also adding missing testcase ID to tracecompiler_whatlog

Bug 1912 - Raptor should take python from the path by default

Minor update

Minor update

helium_7.0-r14027

version 2.12.1 (candidate 1)

release note: 3 fixes for 2.12.1

sf bug 1519: unescaped entities in XML logs (from make errors)

catch up

catch up to default

Merging Bug 1494 fix back in

Release note: Two trace compiler bugs fixed (SF Bug 1494 and one other)

Fixing qt_helloworld test id. Fixing copyright notices.

Renumbering commandline, pdll_arm and toolchain_macro tests to avoid duplicates. Renaming qt_helloworld, commandline and whatcomp to match the test and file names. Allocating ids to qt_helloworld and apply_usecases. Fixing typo in commandline test and enabling it so it actually runs.

Fix winscw trace compile to not use the default UID3 if unspecified

Undoing spelling 'fix'

Removing references to local file system in tc_whatlog test

Oops, typo

Fixing case non-issue just to be tidy (and to save headaches if the tracecompiler ever runs on Linux, possibly)

Fixing annoying stack traces when all you've done is kept a log file open in Vim

Fixing bug 1494 (tracecompiler whatlog) properly this time

Catch up to 2.11.3

Bug 1494 - Trace compiler files not in whatlog - FIX and TESTS though UNTESTED since only run on Linux so far

version 2.12.0 (candidate 3)

catch up to 2.12.0

version 2.12.0 (candidate 2)

merge from FCL

version 2.12.0 (candidate 1)

merge from fix

merge from wip

Merge: installer stuff

Merge

set file format of raptor_meta.py to unix when on unix

cleanups from reviews.

Default embedded Python to 2.6.4 on Windows as on Linux.

Release note: Fix SF bug 1570 - Adding then removing a capability can produce an incorrect result

fix bug 1570: actually *use* the capabilities and fix test for windows and for capability flags.

Merge test + fix for bug 1570

test: check removing then re-adding capabilities. Make sure output exe has the right caps.

fix: bug 1570 - Adding then removing a capability can fail. e.g. ALL -TCB TCB

Correct problems in raptorinstallermaker.py: ensure missing directories' names are printed rather than %s when checking Win32 support directory, and remove final reliance on SBS_HOME environment variable (use only the value passed to the -s switch).

Update Windows installation script to put Python 2.6.4 into the Windows installer.

release: fix for ARM9E + ARMV5 pre-processing confusion

variant platform test

catch up

fix for ARM9E + ARMV5 preprocessing confusion

Test fixes for whatcomp test: Deal with slashes correctly. Use Python 264.

Tab/space problem

Elimitate those testing messups that result from the case of the drive letter in SBS_HOME or EPOCROOT

Use '' to protect backslashes in arguments rather than escaping. For parallel parsing.

In epocroot handling make windows the exceptional case because it is, use os.sep.

filter and wilterwhatcomp cosmetic (style) fixes.

Merge

fix: add .stderr dumfile to log *after* make engine has exited.

Merge

Make TEM targets more unique to avoid problems building armv5 and arm9e

Merge

Merge error fixes with filterwhat stuff

remove fluff

sbs_filter: Fix so that return code is 0 on success (was always 1) filter_whatcomp: add a test for this

Merge filter_whatcomp improvements in.

Protect some parameters from bash, e.g. ";" and "\"

Stderr to a file - avoid xml problems in error messages.

Richard's fix for combined ARM9e/ARMv5 builds

Prevern Path and PATH variables from messing things up

Put stderr into separate log file

Don't mess around with EPOCROOT until actually entering raptor so we know what the original was Put the original epocroot back on the front of the whatcomp output. This allows what output to be either relative or absolute depending on what your epocroot is.

Release notes. Also remember to add FilterCheck

Cause whatcomp output to use the incoming epocroot value. i.e. if epocroot is relative then so is the what output. e.g. if EPOCROOT=\ then the output will be of the form "\epoc32\release\armv5\...." If it's "..\myepocroot" then the output will be "..\myepocroot\epoc32\release\armv5". If it's absolute then the what output will also be absolute.

Strip epocroot off whatcomp filter output.

Release note: sf bug 1558: [Raptor] auto-setting of SBS_HOME by sbs.bat fails when drives differ

Release: DPDEF143051 Raptor doesn't process the LINKEROPTION keyword as Abld

Merge

Test case for Apply linker options