Implemented attributes for class types and properties (based on work by Joost van der Sluis). Added TCustomAttribute - a base class for attributes. Added TUnitInfo record to RTTI. It contains the unit name and unit options (for now only a flag which specifies if the unit contains attributes). Added several tests for attributes.
git-svn-id: trunk@42356 -
Allows compilation of compiler using -dDEBUG_NODE_XML
which will generate a NAME-node-dump.xml file for each
unit, program or library compiled,
containing a XML description of the nodes handled during
compilation of the unit, program or library.
git-svn-id: trunk@42271 -
initialisation/finalisation routines
o this ensures the procdef's mangled name and its first symbol match,
which is required for Darwin when generating debug information and
using ".set" directives to define alias symbols, because a symbol
defined via ".set" cannot be used in a data relocations (and the DWARF
debug generator uses the default mangled name, which used to correspond
to the alias symbol due to the "insert" instead of "append")
git-svn-id: trunk@42094 -
program/library also with -Clflto, all units compiled with that option
will be linked using their bitcode files and LTO
o compiling with -CLflto will compile all units twice: once for normal
(static or smart) linking, and once for LTO. So the result can be
used both with and without LTO.
git-svn-id: branches/debug_eh@41910 -
on pdecsub (node units should not depend on parser units):
o maybe_add_public_default_java_constructor()
o handle_calling_convention()
o create_finalizer_procdef() (replaced with create_outline_procdef())
o insert_record_hidden_paras()
o handle_calling_convention()
o proc_add_definition()
o build_parentfpstruct()
o maybe_guarantee_record_typesym()
o get_first_proc_str()
* factored out the creation of a procinfo for a nested procdef based on a
subnodetree of the current procdef into tprocinfo.create_for_outlining()
git-svn-id: trunk@40773 -
* ppu version update
* -Sf might be used only during system unit compilation
* loadsystemunit factored out so the system unit can be loaded earlier than other units to get features set
git-svn-id: trunk@37708 -
at the end of compiling a unit, also remove it from its owning procsym
in case the procsym does get written to the ppu (possible because you
can have extra overloads in the implementation) (mantis #25283)
o also fixes webtbf/tw4103 on the platforms where this still failed
(on the platforms where it worked, it worked by accedent since the
compiler was accessing memory of a freed procdef)
git-svn-id: trunk@35309 -
Use new cs_checkpointer_called moduleswitches set element.
Use new uf_checkpointer_called PPU flag (reusing obsolete uf_local_browser flag value)
Emit warning if compiled library/program has any code using checkpointer.
* ppu.pas: New constant: uf_checkpointer_called
* globtype.pas: New constant: cs_checkpointer_called
* fppu.pas: Set uf_checkpointer_called flag if cs_checkpointer_called is set in current_settings.module_switches
* i8086/n8086mem.pas: Include cs_checkpointer_called in current_settings.moduleswitches
* ncgmem.pas: Likewise.
* msg/errore.msg: Add new message saying that -gc and -Ur options are incompatible
Add description to -gc option, saying it is experimental.
Add warning at link time for program/library if checkpointer is used in any unit or main code.
* options.pas: if -gc and -Ur options are used, never enable checkpointer code,
instead output a warning that release is incompatible with -gc option.
* pmodules.pas: proc_program: Check all modules for uf_checkpointer_called flag,
emit a warning if checkpointer is used.
* utils/ppuutils/ppudump.pp: Add code for uf_codepointer_called option.
git-svn-id: trunk@34567 -
tasmdata.DefineAsmSymbol() and all routines that call it
o will be used to automatically generate AB_INDIRECT sybols when
necessary
git-svn-id: trunk@34164 -
list, so that the assembler directives generated for Darwin are emitted
correctly for LLVM
o extracted the library init code generation from hlcgobj to pmodules so
we don't add things to another list than what's passed to
gen_proc_symbol_end(), and so it's done where we generate the init
routine for libraries (rather then checking for every routine whether
it's the init routine for a library)
o removed the llvm-specific gen_proc_symbol_end() because the regular one
now does the same
git-svn-id: trunk@34134 -
fmodule.pas, tmodule:
+ new list pendingspecializations which keeps track of all pending specializations of the current module
psub.pas:
* move generate_specialization_procs and related routines to pgenutil
+ new procedure read_proc_body to read a routine's body, cause generate_specialization_procs needs it (unlike the already existing overload in the implementation section, this one can only handle bodies of non-nested routines)
pgenutil.pas:
* generate_specialization_phase2: add the newly specialized generic to the current module's pending specializations
* generate_specialization_procs: reworked so that it uses the new pendingspecializations field instead of walking the global and local symboltable of the current unit
pmodules.pas:
+ add pgenutil to uses due to the moved generate_specialization_procs
+ added test
git-svn-id: trunk@33826 -
Report a warning if a unit is used from an indirectly used package. We are doing this only for the units used in contained units though as in the "contains" section there can't be such units anyway (and just checking all loaded units would lead to false positives).
pkgutil.pas:
+ new procedure check_for_indirect_package_usages() which walks a TLinkedList of used units and warns on every unit that is from an indirectly imported package
pmodules.pas, proc_package:
* when checking all loaded units whether they are from a package or not also check for indirect package usages using the new procedure
........
pkgutil.pas, exportprocsym:
* use a temporary variable for the procdef
* restructure the if-expression a bit to make it better readable
........
pmodules.pas, finish_unit:
* also release the generated proc symbol if the init function isn't needed
........
pkgutil.pas, add_package_libs:
* fix exit condition (on Linux it probably worked by accident :/ )
........
git-svn-id: trunk@33518 -
Keep track whether a package was added directly through a requires section (or the command line) or indirectly.
fpkg.pas, tpackageentry:
+ new field "direct"
pkgutil.pas:
+ add_package: new parameter "direct" which sets the "direct" field of the new package entry
options.pas:
* TOption.interpret_option & read_arguments: adjust call to add_package()
pmodules.pas:
* proc_package: adjust call to add_package()
........
Also load all packages that are required by packages, but that are not part of the directly used packages.
pkgutil.pas, load_packages:
* check the list of required packages for packages that are not yet part of the package list and add these as indirect packages
* establish the links to loaded packages in the required packages once all packages were loaded
........
fpcp.pas, tpcppackage:
+ new method add_required_package() to add a required package while ignoring duplicates
........
pmodules.pas, proc_package:
* add all packages of loaded units as required packages
........
git-svn-id: trunk@33517 -
cclasses.pas, TCmdStrList:
* if doubles are not allowed then do case sensitive checks instead of insensitive ones
........
Correctly handle export/import generation depending on whether we're compiling for a system that needs indirect imports or not.
systems.pas:
* replace systems_package_need_exports with systems_indirect_var_imports
pmodules.pas, proc_package:
- remove check for systems_package_need_exports
pkgutil.pas:
* insert_export, export_unit: export the direct or the indirect variable symbol depending on systems_indirect_var_imports
........
ncgrtti.pas, TRTTIWriter.write_extra_rttisyms:
* enumdef_rtti_string2ordindex & enumdef_rtti_ord2stringindex: put the indirect suffix at the end of the symbol
........
pkgutil.pas, add_package_libs:
* check for systems_indirect_var_imports instead of systems_package_needs_exports
........
expunix.pas, texportlibunix:
* generatelib: check all procdefs whether they have the alias and not only the first one
........
pkgutil.pas:
* insert_export: directly report the symbol type instead of its ordinal to ease finding of problems
........
pkgutil.pas, export_procsym:
* restructure if for better readability
+ also check the procdef for po_has_public_name
........
Ensure that all implicitely imported units are indeed part of the package.
pmodules.pas, proc_package:
* instead of checking the units loaded by the current module we need to check *all* units that are loaded for the package
........
Fix inclusion of PPUs that don't have a corresponding object file.
pkgutil.pas, RewritePPU:
* don't abort if a unit does have the flag uf_no_link set
* only abort if both uf_static_link and uf_no_link are not set
........
Generate an error if a unit listed as in "contains" is part of a directly or indirectly required package.
pmodules.pas, proc_package:
* walk all units and check whether it had been loaded from a package, but is also part of the units of the current module; if so then generate an error
........
git-svn-id: trunk@33516 -
Correctly parse the directives DenyPackageUnit and WeakPackageUnit
ppu.pas:
+ add flags uf_packagedeny and uf_packageweak
scandir.pas:
+ new procedure do_moduleflagswitch() which parses a ON/OFF/+/- argument and sets or clears a flag in the current module
+ new procedure dir_denypackageunit which handles DenyPackageUnit
* implement dir_weakpackageunit (and move to the correct location ;) )
* InitScannerDirectives: add dir_denypackageunit handler
........
Respect DenyPackageUnit flag.
pmodules.pas, proc_package:
* check all contained units that are not already part of a package for their uf_package_deny flag and report an error for each that has it set
........
Do not check whether all units are used as by definition all units of a package are considered as used.
pmodules.pas, proc_package:
- remove call to current_module.allunitsused
........
Check whether a unit has been implicitely imported in a package. A unit is considered as implicitely imported if it is not part of a required package nor part of the units listed in the contains section. This note is useful (Delphi even provides a dialog in that case) as a package with implicitely imported units /might/ become incompatible with other packages (e.g. if another package includes that unit uses that package and includes that unit explicitely; of course that is the same as if both package included it explicitely, but with the hint one knows where to look).
pmodules.pas, proc_package:
* while walking the loaded units also check whether any of them not contained in a package was part of the contained units which are the same as the current module's used units
........
Generate CRC for package files
pcp.pas, tpcpfile:
+ new field do_crc which controls CRC generation
+ override putdata() method to generate CRC when data is written
* resetfile: enable do_crc by default
........fppu.pas, tppumodule:
* loadfrompackage: mention if a unit is loaded from a package
........
fpkg.pas, tcontainedunit:
+ new fields offset and size for the PPU data stored inside the PCP
fpcp.pas, tpcppackage:
* readcontainedunits & addunit: correctly initialize offset and size to 0
........
Store the modified PPU files directly inside the PCP and thus get finally rid of the .ppl.ppu files.
entfile.pas:
+ new entry type ibpputable
pkgutil.pas:
* adjust RewritePPU to work on a stream as output instead of a filename
fpcp.pas, tpcppackage:
+ new method writepputable() which writes the offsets and sizes of all contained units (not part of CRC!)
+ new method writeppudata() which rewrites all contained PPUs directly into the PCP after the ibend entry (Note: the data is written 16 Byte aligned to ease viewing of the PCP and its contained PPUs in a hex editor)
+ new method readpputable() which reads the offsets and sizes of all contained units
+ new method getmodulestream() which returns a substream for a contained module
* loadpcp: also call readpputable()
* writepcp: first write an empty pputable, then finish writing all data that requires the put*/write* methods of the pcpfile, then use writeppudata() to write all PPUs and finally write the correct pputable at the original location
fppu.pas, tppumodule:
* loadfrompackage: don't read the PPU from a file if it is contained in a package, but using the new tpcppackage.getmodulestream() and tppumodule.openppustream() methods
pmodules.pas, proc_package:
* don't rewrite the PPUs here
pcp.pas:
* increase CurrentPCPVersion
........
Fix cycling
........
git-svn-id: trunk@33514 -
pmodules.pas, proc_package:
* generate the dummy entry symbol only on Windows systems (ToDo for other systems that might need it)
........
options.pas:
* TOption.interpret_option & read_arguments: use add_package() from pkgutils instead of addpackage() from fpkg
fpkg.pas:
- remove no longer needed addpackage()
........
pkgutil.pas:
+ new procedure add_package_libs() to add all packages as libraries that are also linked against
pmodules.pas:
* proc_package & proc_program: add all package libraries to the linker before creating the final binary
........
pkgutil.pas, add_package:
* package names are case insensitive so add them in uppercase to correctly detect duplicates
........
Keep track of how many units are used from each required/requested package and only link those from which any units are used.
fpkg.pas, tpackageentry:
+ new field usedunits to count how many units of the package are used
pkgutil.pas:
+ new procedure add_package_unit_ref() to increase the amount of used units for the given package
pmodules.pas:
* proc_package & proc_program: check which units of all loaded ones are provided by other packages
........
git-svn-id: trunk@33510 -
Don't try to export generic symbols.
pkgutil.pas:
* exportabstractrecordsymproc: do not export generic type symbols or their nested symbols
........
Ensure that the correct name is used for the PCP file.
proc_package:
* use current_module.modulename instead of module_name for the base filename of the PCP file
........
Add a possibility to track if a unit was loaded from a package.
fmodule.pas, tmodule:
+ new field "package" which is Nil if the unit is not (yet) part of a package or a reference to the package this unit was loaded from
fppu.pas, tppumodule:
* loadfrompackage: set the "package" field of the module to the package it's loaded from.
........
Ensure that units loaded from a package are not recompiled if the source files should happen to be available.
fppu.pas, tppumodule:
loadppu: set the state to ms_compiled instead of ms_load if the unit was loaded from a package (additionally close the PPU file as it's no longer needed)
........
Add support for reading/writing required packages from/to the PCP file
fpcp.pas, tpcppackage:
+ new methods writerequiredpackages and readrequiredpackages
* loadpcp: use readrequiredpackages
* savepcp: use writerequiredpackages
pcp.pas:
* increase PCP version
........
When compiling a package, handle only those units that are not yet part of a package.
pmodules.pas, proc_package:
* only export units that are part of the package
* only add units to the package if they are really part of the package
* don't rewrite the PPU if the unit is not part of the package
* don't link the unit's files if it is not part of the package
........
Add support for parsing required packages.
pkgutil.pas:
+ new function add_package to add a package to the list of available packages with the possibility to check for duplicates
* load_packages: also load all required packages
pmodules.pas, proc_package:
* create the tpcppackage instance earlier (and use the module name as read from the source file as package name)
* clear the list of packages in case the user passed any using -FPxxx
* parse the "requires" section like a list of units and add each full identifier as a package to load
* before parsing the "contains" section load all packages, so that all units can be correctly resolved
........
Correctly create import libraries for packages as well.
pkgutil.pas, createimportlibfromexternals:
* instead of processing units without the uf_in_library flag, only process those that don't have a package reference set (thus becoming part of the program/library or the package)
- remove unneeded "pkg" parameter
pmodules.pas:
+ proc_package: create the import library if the package requires other packages
* proc_program: adjust call to createimportlibfromexternals
........
Ensure that the reference to the System unit is correctly set up for packages. Among other things this is needed to compile a package with the -gl option (though debug information doesn't work yet).
pmodules.pas:
* convert AddUnit to a function and let it return the newly loaded module
* proc_package: when parsing the contained units ensure that we correctly set up the System unit reference if we are to contain the System unit
* proc_package: also set up the System unit reference once all units are loaded and this hasn't happened yet (because all contained units are already compiled)
........
git-svn-id: trunk@33502 -
Generate the import library for each used package.
pkgutil.pas:
- remove function createimportlibfromexports
+ new function createimportlibfromexternals which uses the imported unit symbols of a unit to generate the correct import library for each used package used in a program or library
pmodules.pas, proc_program:
* move the generation of the import library further down until all tables were generated of which entries might reside in a package
* use createimportlibfromexternals to generate the import library for the program/library
-- Diese und die folgenden Zeilen werden ignoriert --
M pmodules.pas
M pkgutil.pas
git-svn-id: trunk@33493 -
Merged revision(s) 28796, 28837-28845, 28847-28850, 28852, 32135 from branches/svenbarth/packages:
Provide possibility to pass packages and search paths for packages as parameters.
fpkg.pas:
+ new unit which contains the base types related to package files (most importantly "tpackage")
globals.pas:
+ new variable "packagesearchpath" which contains all paths in which package files should be looked for
+ new variable "packagelist" which contains a list of all packages that should be used in a program or library
* InitGlobals & DoneGlobals: initialize/finalize "packagesearchpath" accordingly ("packagelist" is handled in unit fpkg using a init/done-callback)
options.pas:
+ TOption: new fields "parapackagepath" and "parapackages" to keep track of package search paths and package files passed as parameters
* TOption.interpret_option: use '-Fp' for package search paths and '-FP' for package files
* read_arguments: apply the passed package search paths and packages to their respective containers
........
+ add a new unit which will contain basic functions related to handling packages
........
Move package related functions from pmodules to pkgutil.
pmodules.pas => pkgutil.pas:
* createimportlibfromexports
* varexport
* procexport
* insert_export
* RewritePPU
........
* adjust indentation
........
Extract the code to export the symbols of a unit to its own function in pkgutil so that less functions need to be exported.
pmodules.pas, proc_package:
* move code to export the symbols of a unit to new function export_unit
pkgutil.pas:
+ new function export_unit
- remove exports of procexport, varexport and insert_export
........
Some small fixes for package parsing.
pmodules.pas, proc_package:
* use orgpattern instead of pattern to build the module name (like is done in uses sections)
* ignore duplicates when generating exports
- no need to generate an import library for the package; that is done by the program/library that uses the package
........
+ new unit fpcp of which the class tpcppackage handles the reading and writing of package metadata from/to pcp files (equivalant to tppumodule).
........
+ add unit which contains representation of a PCP file (tpcpfile) like tppufile is for units.
........
Improve export generation.
pkgutil.pas:
+ new function exportprocsym to correctly export a procedure with all its aliases
+ new function exportabstractrecordsymproc to export the members of structured types
* insert_export: handle also namespacesym and propertsym (by ignoring them)
* insert_export: correctly export classes, record and objects
* insert_export: use new exportprocsym function to export a procsym
* insert_export: only export public variables of a static symtable
........
+ add entry constants for the name of the package and the package file names, both used by a PCP file
........
* use messages to get rid of most writelns related to package loading
........
Add additional entry types for PCP files
entfile.pas:
+ new entries ibstartrequireds and ibendrequireds to store the list of required packages
+ new entries ibstartcontained and ibendcontained to store the list of contained units
+ new entries ibstartppus and ibendppus to store the list of contained PPU files
........
Generate the PCP file once the package file and the used units were compiled correctly.
pmodules.pas:
* proc_package: generate the PCP file upon successful compilation
........
Add the possibility to load all packages supplied as parameters.
pkgutil.pas:
+ new function load_packages to load all packages supplied as parameters
pmodules.pas, proc_program:
* use load_packages to load all packages before any unit is loaded
........
Add code which tries to load a unit from a package first and only then as usual.
fppu.pas, tppumodule:
+ new method loadfrompackage which searches all available packages for the unit and loads it from there if found
* loadppu: first try to load the unit from a package if any are available
........
Don't link objects files of a unit that is provided by a package.
pmodules.pas, proc_program:
* if a unit has uf_in_library set we must not include it in the units we link against
........
git-svn-id: trunk@33452 -
Merged revision(s) 28793, 28903, 31951 from branches/svenbarth/packages:
psub.pas, read_proc:
* don't generate assembler symbols for compiler internal code
........
Ensure that unneeded initialization/finalization procedures don't leave assembler symbols.
ncgutil.pas:
+ new function release_proc_symbol which removes all assembler symbols of a procdef
pmodules.pas, proc_unit:
* if the initialization or finalization procedure is empty also remove their assembler symbols so they are exported
........
Ensure that readonly sections that need relocations are indeed declared as such.
Thanks to Jonas for this.
ncgvmt.pas, TVMTWriter:
* writeinterfaceids: use sec_rodata instead of sec_rodata_norel
........
git-svn-id: trunk@32982 -
Extract reading of the PPU/PCP version into tentryfile.
entfile.pas:
+ add new getversion method which uses the pointer returned by getheaderaddr to read the PPU/PCP version
ppu.pas:
- remove GetPPUVersion method
fppu.pas, pmodules.pas, utils/ppufiles.pp, utils/ppumove.pp, utils/ppuutils/ppudump.pp
* replace call to GetPPUVersion by call to getversion
........
git-svn-id: trunk@32980 -
Add a new target flag which determines whether the target supports dynamic packages or not.
systems.pas:
+ new flag tf_supports_packages
pmodules.pas, proc_package:
* error out when flag tf_supports_packages is not set for the current target
msg/errore.msg:
+ error message for when tf_supports_packages is missing and a package file is compiled
........
git-svn-id: trunk@32978 -
Extract functionality that is shared between the metadata files for units (PPU) and for packages (PCP) into a parent class called tentryfile
+ add new unit entfile which contains the new tentryfile class and related types and constants
* ppu.pas:
- remove methods, fields, types and constants which were moved to entfile.pas
* replace the parts of tppuheader shared with tentryheader by a field of type tentryheader
fppu.pas, pmodules.pas, utils/ppumove.pp, utils/ppuutils/ppudump.pp:
+ add entfile to uses
* adjust access to common header fields
node.pas, symdef.pas, symsym.pas, symtable.pas, wpoinfo.pas, utils/ppufiles.pp:
+ add entfile to uses
........
git-svn-id: trunk@32976 -
"don't free even if not registered"; use for defs that may not be written
to a ppu file, but that must nevertheless survive the compilation of the
current module
* mark all defs created for para locations as "don't free even if not
registered", because we don't discard and recalculate all para locations
after a module has been compiled (since that's not needed)
o solves issues if the paralocations for a routine in the interface of
unit A are calculated while the implementation of unit B gets
compiled, and a new reusable type is allocated at that point which
is not used anywhere else (after r32160)
git-svn-id: trunk@32235 -
o this allows us to write the VMTs earlier (before the synthetic method
implementations are generated), which means we can create new synthetic
methods while generating the VMTs (for interface trampolines)
git-svn-id: trunk@31635 -
* for now all objectdefs are created as registered
Note: an additional parameter instead of an overload is used for tobjectdef.create as otherwise both constructors would need to be overridden in potential descendant CPU-specific classes...
git-svn-id: trunk@31592 -
* for now all typesyms are created as registered
Note: an additional parameter instead of an overload is used for ttypesym.create as otherwise both constructors would need to be overridden in potential descendant CPU-specific classes...
git-svn-id: trunk@31591 -
o blocks are implemented as a variation of procedure variables
o declaration of a block variable: "test: procedure(c: char) is block;"
(C equivalent: (void)(^test)(char c) )
o the compiler automatically converts procedures/functions whose address
is passed to a block parameter or assigned to a block variable into
a "block". This consists of
1) generating a block descriptor (containing the size of the "block
literal" (see below) and the signature of the invocation function
encoded as an Objective-C selector)
2) generating a wrapper function around the original funcion (with C
calling convention), that has an extra first hidden parameter
(marked as vo_is_parentfp in the compiler) whose type is a pointer
to the describing "block literal"
3) generating the "block literal", which contains a pointer to an
external variable indicating whether this block captures context or
not, some flags (see compiler/blockutl.get_block_literal_flags for
info), a pointer to the wrapper function and a pointer to the
descriptor. In the future, it will also contain captured variables.
o right now, only global procedures/functions can be converted to blocks
(because they don't require state capturing). The next steps are (Object
Pascal) methods (not Objective-C methods, because Objective-C method
procvars don't exist) and finally nested functions
o on Mac OS X, the functionality will only work on Mac OS X 10.7 and later,
because we have to use the so-called "ABI.2010.3.16" to ensure that
our blocks aren't called as variadic functions by the runtime (which
came out after the Mac OS X 10.6 release)
o while the currently implemented functionality does not require any
library support at all, there's no use enabling it on other platforms
because unless it has been confirmed to work with a blocks runtime,
there's no point in using blocks (they're just somewhat bulky procvars
right now). Enabling it on other platforms (in combination with the
GNUStep Objective-C run time), should simply be a matter of adding
the right {$linklib xxx} statement to rtl/inc/blockrtl.pp file, adding
that file to Makefile.fpc for that platform and adding that platform
to the compiler/systems.systems_blocks_supported set
git-svn-id: branches/blocks@28232 -
o made all (non-abstract) tdef and tsym constructors virtual
o added c*def/c*sym classref types for every (non-abstract) t*def/t*sym
class
o added cpusym unit for every architecture that derives a tcpu*def/tcpu*sym
class from the base classes, and initialises the c*def/c*sym classes with
them. This is done so that the llvm target will be able to derive from
the tcpu*def/sym classes without umpteen ifdefs, and it also means that
the WPO can devirtualise everything because the c* variables are only
initialised with one class type
o replaced all t*def/t*sym constructor calls with c*def/c*sym constructor
calls
git-svn-id: trunk@27361 -
This improves compiling speed a bit (two iterations over symtables replaced by one, code generator is created once per unit rather than once per class).
In perspective it makes possible to reduce amount of generated smartlink sections and global labels.
git-svn-id: trunk@24269 -
The cause of the internal error was the following:
We have a generic in an unit ("A") which uses another unit ("B") in the implementation section and this other unit uses unit A in the interface section. Now the generic is specialized in the interface section of B. This leads to the problem that in unit A when it tries to load the globalsymtable of unit B that globalsymtable will be Nil, because parsing of the interface section is not yet finished. Thus the change in pgenutil.pas, specialization_init: if the unit is still "in_interface" the localsymtable needs to be used instead of the globalsymtable.
This doesn't necessarily lead to a compiling test though, as there is the following possibility:
Unit A contains a generic class/record (with methods) and uses unit B in the implementation section. This unit B also contains a generic class/record (with methods) and uses unit A in the implementation section. Both units contain a specialization of the other unit's generic outside of it's own generics (such that generate_specialization is fully triggered). Let's assume compilation starts with unit A and we reach the uses of unit B. Now compilation switches to unit B and completes as unit A is already registered and in compilation. The problem now is that the generic in unit A still contains unresolved forward declarations as the implementation section of A was not yet parsed which will lead to "forward declaration not solved" errors (Note: Delphi compiles this).
The solution to this is the following: if a generic is specialized from another unit which is not in state ms_compiled then the unit of the specialization needs to wait for the unit of the generic. So the specialization's unit adds itself into a list of waiting units of the generic's unit. Now inside "proc_unit" we need to check whether this module is waiting for other modules and if so avoid "finishing" the unit (which means generating the methods of the specialization, generating assembler code and ultimately freeing the scanner and PPU). Now when the generic's unit finishes we need to check whether other modules are waiting for it and finish them (of course it's a bit more complicated in reality, but that pretty much sums it up).
+ globstat.pas: Added an unit which handles the saving and restoring of the global state which was originally inside "parser.pas, compile" so that Don't Repeat Yourself (DRY) is respected.
* fmodule.pas, tmodule:
+ add fields to keep track of the units the module is waiting for and which modules are waiting for the module
+ add field for the saved global state (raw pointer to avoid circles)
+ add field for the state which is needed to finish the unit (raw pointer to avoid circles)
+ move the code which was used in "parser.pas, compile" after a module was successfully compiled to the new virtual method "end_of_parsing"
+ fppu.pas, tppumodule.end_of_parsing:
free the ppufile here
* pgenutil.pas:
+ add new procedure "maybe_add_waiting_unit" which adds the specialization's unit to the waiting list of the generic if that unit is not yet compiled
* generate_specialization: call the new function when we add a new (true) specialization
* specialization_init: instead of not adding implementation units at all check whether the unit is still parsing the interface section and add the localsymtable in that case
* pmodules.pas:
* change "proc_unit" to a function which returns "true" if the unit was already finished (no need to wait for other units)
+ move the code from "proc_unit" from "generate_specialization_procs" on to a new procedure "finish_unit" which
* this procedure is either called immediately in "proc_unit" if the unit does not need to wait for other units or from "finish_unit" itself if a unit that is waiting for the given unit does no longer wait for another module (special care is taken in proc_unit to avoid circles)
* parser.pas, compile:
* correctly handle the case if an unit is not finished
* use the new global state functionality from globstat.pas
* pay special attention when calling "set_current_module" (see comment at that call)
+ add tests from 22160
+ add test for above mentioned "diamond" case
git-svn-id: trunk@22452 -
