This variable can be used to build XmlAda with a given RTS,
and it also adjust the installation directory where XmlAda
gets installed afterwards.
For Q802-040.
Change-Id: I54b18f184517c7eded0409c264ce8030f33d6782
When trying to build XmlAda for e500v2-vx6, one calls configure
with --target=e500v2-wrs-vxworks, which results in the Makefile
containing the following definition for TARGET:
TARGET=powerpc-wrs-vxworks
This is because the --target has been canonicalized by config.sub,
and this is expected. However, this causes problems during the build,
since we then call gprbuild with --target=powerpc-wrs-vxworks
instead of the target we specified during the configure.
This patch fixes the issue by declaring a new Makefile variable
called TARGET_ALIAS, and using that for gprbuild's --target=
option, as well as the installation prefix.
Unveiled while working on O611-016.
Change-Id: I7cf3b24c48d26af0930813b1aff304dc78561aad
For the purpose of sanity checking xmlada on VxWorks (and possibly
other cross platforms), add cross-comilable examples, one per dom, sax
and schema, that have all xml/xsd sources embedded in the test's main.
Add new cross- and existing examples to the installation of xmlada.
Change-Id: I56848955171a3b1f16f38bc3ecd7cef253af6346
Fix examples to import proper project names,
fix domschemaexample to not unset the namespeces feature,
add examples to the testing.
Change-Id: Ic5ba519f9ae80f4df3931757a1fcdb9b72e4f540
When a XSD file is referenced via a relative path with symbolic links
or "." and "..", it could be parsed several times, possibly resulting
in errors like duplicate IDs.
P801-022
Change-Id: I7e12b67e3018b3fc3aeeccb5c428c9aea14dfa39
Needed on some platforms like Darwin for shared libs that might get
repositioned by installers.
Also useful for passing -d to monitor progress
Requested on githuba for commit 0f5b329
Change-Id: Icf78b82cd9f7bbcf81b9d3bbb11809078f4c84f1
If a <choice> has two or more elements with a maxOccurs property,
they will allow a mix of the elements, when it should only allow a
repeat of one of the elements.
The simple fix to the state machine is in schema-schema_readers.adb.
But this change resulted in a regression in the testsuite, when one
of the elements in the <choice> is a <any>. At this point, we are
trying to replace the states in the state machine, but did not
properly chain them, and therefore some states accessible via
empty transitions were not properly accessible.
The order of suggested elements in error messages was reversed in
some cases, but the error message is still value, so adjust the
baseline.
P621-020
Change-Id: I1cfb58e2fd45cfd0fc1660fa8727ab679543ba5b
