references to defs that are not stored in the ppu files (causing
errors) and are never used for inlining (only the specialisations
may be inlined) (fixes webtbs/tw27658.pp after r32031)
git-svn-id: trunk@32050 -
don't need a separate field for this
o changed the order in which symid and realname are stored to/loaded from
the ppufile (because tsym.create now initialised tsym.symid, and it's
called from ppuload) -> increased ppu version
git-svn-id: trunk@32031 -
It fixes PIC code generation with GOT for i386 with enabled optimizations. Bugs #28667, #28668.
Prior the fix I have not been able to compile even RTL with -O2 due to not enough free registers, since EBX is reserved for GOT.
It can be further optimized to teach register allocator to not spill the GOT register if possible.
git-svn-id: trunk@32020 -
The bug has been triggered only with the following conditions:
- tf_safecall_exceptions, tf_safecall_clearstack in target flags;
- fixed stack is used.
Since only the i386-android target meets these conditions atm, the bug has been unnoticed.
git-svn-id: trunk@32005 -
ptype.pas, resolve_forward_types:
* in case of the def of the forward being the same as the genericdef of the def the forwarddef belongs to the typesym needs to be changed to the one of the specialization
tests/webtbs/tw28674.pp:
* adjusted text
git-svn-id: trunk@31987 -
ptype.pas, resolve_forward_types:
* fix an access violation in case of inline pointer declarations
* fix the check whether it's a valid use of a generic (parse_generic is useless here...)
+ added test
git-svn-id: trunk@31910 -
pexpr.pas:
* initialize srsym in the classrefdef case; it should not have let to problems, because it's only used if erroroutp1 is false which is only the case if the symbol is assigned, but better save then sorry...
git-svn-id: trunk@31887 -
resourcestring data .rsj files in case the source file is interpreted as
UTF-8. Previously, the individual UTF-8 bytes were each stored in a
separate widechar in the Json file (mantis #28717)
* due to the fact that rstconv didn't use the cwstring unit on Unix, rstconv
until now just concatenated the bytes stored in the widechars of the Json
file on those platforms, i.e., the strings put in the resource file were
byte for byte equal to what was in the source file. On Windows, these bytes
were interpreted as individual widechars, converted to the
DefaultSystemCodePage and then written. This means that for anything but
ISO-8859-1 (where every widechar from #0000 to #0255 maps to #0 to #255),
the output got corrupted.
In order to keep compatibility with the old behaviour whereby rstconv wrote
the resource strings using the same encoding as in the source file (except
if the data got completely corrupted, in which case compatibility is
useless), we now store all resourcestrings twice in the .rsj file: once as
the exact byte sequence from the source file, and once (properly) encoded
in UTF-16.
By default, rstconv will use the byte string and just write that one to the
resource file. Additionally, there is a new -p option that accepts a code
page name (see rstconv -h for the list of supported names), which can be
used to make rstconv use the UTF-16 version and convert that to the desired
code page (as long as the system on which rstconv runs supports that
codepage).
And this also finally resolves mantis #6477.
git-svn-id: trunk@31881 -
* postfixoperators: generate specialize nodes in mode Delphi
* factor: don't call postfixoperators() if returned node is a specialize node
* sub_expr: handle specialize nodes
* sub_expr: extract common code handling specializations in < and as/is into a nested function
git-svn-id: trunk@31859 -
* extend handle_specialize_inline_specialization() with the ability to handle generic functions
* have handle_factor_typenode() and postfixoperators() make use of this extension
git-svn-id: trunk@31856 -
actually export them from the .exe or .dll, but they can now be used as
callbacks)
+ support win16 smart callbacks via the $K switch
git-svn-id: trunk@31835 -
compiler (BP7 compatibility) instead of controlling stack frame elimination
(which isn't supported by the compiler on i8086 anyway, due to i8086's lack
of support for stack relative addressing)
git-svn-id: trunk@31834 -
when {$modeswitch systemcodepage} gets disabled without having been
enabled first (such as when initially setting the syntax mode)
(mantis #28718)
git-svn-id: trunk@31831 -
for all non-ppc(32/64) Darwin platforms
o pass the macosx-version-min/iphoneos-version-min to clang as an assembler,
so that it properly sets this information starting with Xcode 7 (solves
errors when targeting the iOS simulator, and warnings about object files
being compiled for a different OS X version when targeting (Mac) OS X)
o the old assembler is still selectable via -Aas-darwin (required with
Xcode 3.1.x and older)
o since the first Xcode version that shipped with Clang is Xcode 3.2, which
is available for Mac OS X 10.6, most users should not encounter any issues
with the new default (in fact, it fixes some tests for x86 because Clang
supports some instructions that "as" doesn't). Clang does not support
Stabs however, so -gs does require the use of -Aas-darwin
git-svn-id: trunk@31830 -
GNU as from binutils for Darwin, and Apple's "as" is based on an old
version of GNU as)
o this will not cause any backward compatibility problems, as the
previous identifier for as_darwin was "AS-Darwin" and the compiler
compared the uppercase value of the -A parameter to the identifier,
so it was not explicitly selectable earlier. The new name is "AS",
so it is explicitly selectable via -Aas like on other platforms.
git-svn-id: trunk@31806 -
oldcctools are no longer shipped with Xcode 7 and Clang assembler doesn't
support Stabs (and DWARF2 has been working better than Stabs since a long
time already)
git-svn-id: trunk@31774 -
+ export parse_proc_dec_finish() from the unit
pgenutil.pas:
* adjust generate_specialization_phase2() to correctly specialize procdefs
git-svn-id: trunk@31772 -
* extend parse_proc_head() with the ability to deal with specializations of procdefs
Note: such a procdef is created as unregistered as this is used to determine the correct overload in tcallcandidates (through the use of the yet to be adapted generate_specialization_phase2())
git-svn-id: trunk@31771 -
* extend parse_proc_head() with support for /parsing/ generic functions (at least in mode Delphi, mode ObjFPC depends on the new isgeneric parameter to be set)
* adjust parsing of interface mappings with a generic interface (note: in mode ObjFPC this now requires a "specialize" directly before the generic interface's name, which is more in line with other uses of "specialize")
pexpr.pas, factor:
* don't call postfixoperators() if hadspecialize is set
tests/test/tgeneric79.pp:
* adjust test to changed syntax
git-svn-id: trunk@31769 -
* extend do_member_read() with the possibility to pass a specialization context on to do_proc_call() if the symbol is a procsym
git-svn-id: trunk@31767 -
* do_proc_call can take a tspecializationcontext and pass it on to ccallnode.create()
Note: pointers to generic functions/methods are not yet supported, so there's a temporary, non-translated error message for these constructs
git-svn-id: trunk@31766 -
+ add error message if a generic method is declared inside an interface
+ add error message if a generic method is declared as virtual
git-svn-id: trunk@31764 -
* extend tcallnode with the ability to pass a tspecializationcontext so that tcallcandidates can do a final specialization
* the final procdef is registered at the end of tcallnode.pass_typecheck
git-svn-id: trunk@31763 -
+ new token buffer that holds the declaration of the generic function/method while the one from tdef contains the body
ppu.pas:
* increase PPU version
utils/ppuutils/ppudump.pp:
* also read the declaration token buffer, it's however not printed yet (ToDo!)
git-svn-id: trunk@31759 -
* no longer create the type symbol and the defs as registered
pdecl.pas, types_dec:
* register the syms and defs of the parsed generic parameters
git-svn-id: trunk@31758 -
* for now all undefineddefs are created as registered
Note: an additional parameter instead of an overload is used for tundefineddef.create as otherwise both constructors would need to be overridden in potential descendant CPU-specific classes...
git-svn-id: trunk@31757 -
