This is probably not needed, but it is probably safer to make sure the
old files are removed instead of watching the core rebuild endlessly
when its not.
In libretro-buildbot-recipe.sh line 148:
if [ -z "${HELPER} ${MAKE}" ]; then
^-- SC2157: Argument to -z is always false due to literal strings.
In libretro-buildbot-recipe.sh line 348:
echo "CLEANUP CMD: $@" 2>&1 | tee -a "$LOGFILE"
^-- SC2145: Argument mixes string and array. Use * or separate argument.
In libretro-buildbot-recipe.sh line 352:
echo "CLEANUP CMD: $@" 2>&1 | tee -a "$LOGFILE"
^-- SC2145: Argument mixes string and array. Use * or separate argument.
In libretro-buildbot-recipe.sh line 372:
echo "BUILD CMD: ${CMAKE} $@" 2>&1 | tee -a "$LOGFILE"
^-- SC2145: Argument mixes string and array. Use * or separate argument.
In libretro-buildbot-recipe.sh line 380:
echo "BUILD CMD: $@" 2>&1 | tee -a "$LOGFILE"
^-- SC2145: Argument mixes string and array. Use * or separate argument.
In libretro-buildbot-recipe.sh line 388:
echo "BUILD CMD: $@" 2>&1 | tee -a "$LOGFILE"
^-- SC2145: Argument mixes string and array. Use * or separate argument.
In libretro-buildbot-recipe.sh line 480:
echo "BUILD CMD: $@" 2>&1 | tee -a "$LOGFILE"
^-- SC2145: Argument mixes string and array. Use * or separate argument.
In libretro-buildbot-recipe.sh line 626:
if git config --file "$DIR/.gitmodules" --name-only --get-regexp path 2>&1 >/dev/null; then
^-- SC2069: The order of the 2>&1 and the redirect matters. The 2>&1 has to be last.
In libretro-buildbot-recipe.sh line 1018:
if [ -n ${CUSTOM_BUILD} ]; then
^-- SC2070: -n doesn't work with unquoted arguments. Quote or use [[ ]].
In libretro-buildbot-recipe.sh line 1045:
if [ -n ${CUSTOM_BUILD_DEBUG} ]; then
^-- SC2070: -n doesn't work with unquoted arguments. Quote or use [[ ]].
This allows each core to define which archs it builds for.
Currently the scripts force build all archs even for cores that
cannot support them, ala dolphin on 32 bit archs.
This checks for the STRIPPATH varible in the android path. Only
setting that variable for the android cmake targets. JNI outputs
already stripped shared objects.
A few cores built with the LEIRADEL command had an issue where a
hack used to build certain cores broke and caused them to be built
every time the buildbot encountered the recipe.
To fix this the build_libretro_leiradel_makefile function was removed
and incorporated into the build_libretro_generic_makefile function.
This allowed the script to easily use new logic to build multiple cores
with one recipe which entirely avoids the need for the broken hack.
The script will no longer git pull, but will fetch the any changes
from the remote branch and then reset the repo to that branch.
If there are changes the HEAD commit will change and it will rebuild.
This will also work if the old history was changed instead of the new
history.
During one of my clean up efforts I neglected to make sure the quotes
around '-G "MYSYS Makefiles"' were not evaluated. Escaping the variable
containing this option works, but should be noted it only works with bash.
Since this script is inherently a bash script that should be no problem.
Fixes https://github.com/libretro/libretro-super/issues/675
This could be useful for travis to capture regressions if the only
one of the cores is affected. Additionally it is required for fixing
travis with mame2014. If the core is not bsnes-libretro, bsnes-mercury
or mame2014 the $CORE variable in the travis scripts will be ignored.
They now use different source repos so this is not needed.
In the event this is no longer true there are now better
ways to do this, see how bsnes is handled where it loops
over the desired targets.
If the checked out branch does not match the branch name in the recipe
file the script will now make a shallow fetch of the correct branch,
checkout the new branch and then remove the old one. If this is done it
will also build a core with the new branch.
Francisco Javier Trujillo Mata