+=item 4
+
+Checks C<atime> and C<mtime> of C<stat()> - unfortunately, HPFS
+provides only 2sec time granularity (for compatibility with FAT?).
+
+=back
+
+=back
+
+=head2 Installing the built perl
+
+If you haven't yet moved C<perl*.dll> onto LIBPATH, do it now.
+
+Run
+
+ make install
+
+It would put the generated files into needed locations. Manually put
+F<perl.exe>, F<perl__.exe> and F<perl___.exe> to a location on your
+PATH, F<perl.dll> to a location on your LIBPATH.
+
+Run
+
+ make installcmd INSTALLCMDDIR=d:/ir/on/path
+
+to convert perl utilities to F<.cmd> files and put them on
+PATH. You need to put F<.EXE>-utilities on path manually. They are
+installed in C<$prefix/bin>, here C<$prefix> is what you gave to
+F<Configure>, see L<Making>.
+
+If you use C<man>, either move the installed F<*/man/> directories to
+your C<MANPATH>, or modify C<MANPATH> to match the location. (One
+could have avoided this by providing a correct C<manpath> option to
+F<./Configure>, or editing F<./config.sh> between configuring and
+making steps.)
+
+=head2 C<a.out>-style build
+
+Proceed as above, but make F<perl_.exe> (see L<"perl_.exe">) by
+
+ make perl_
+
+test and install by
+
+ make aout_test
+ make aout_install
+
+Manually put F<perl_.exe> to a location on your PATH.
+
+B<Note.> The build process for C<perl_> I<does not know> about all the
+dependencies, so you should make sure that anything is up-to-date,
+say, by doing
+
+ make perl_dll
+
+first.
+
+=head1 Building a binary distribution
+
+[This section provides a short overview only...]
+
+Building should proceed differently depending on whether the version of perl
+you install is already present and used on your system, or is a new version
+not yet used. The description below assumes that the version is new, so
+installing its DLLs and F<.pm> files will not disrupt the operation of your
+system even if some intermediate steps are not yet fully working.
+
+The other cases require a little bit more convoluted procedures. Below I
+suppose that the current version of Perl is C<5.8.2>, so the executables are
+named accordingly.
+
+=over
+
+=item 1.
+
+Fully build and test the Perl distribution. Make sure that no tests are
+failing with C<test> and C<aout_test> targets; fix the bugs in Perl and
+the Perl test suite detected by these tests. Make sure that C<all_test>
+make target runs as clean as possible. Check that C<os2/perlrexx.cmd>
+runs fine.
+
+=item 2.
+
+Fully install Perl, including C<installcmd> target. Copy the generated DLLs
+to C<LIBPATH>; copy the numbered Perl executables (as in F<perl5.8.2.exe>)
+to C<PATH>; copy C<perl_.exe> to C<PATH> as C<perl_5.8.2.exe>. Think whether
+you need backward-compatibility DLLs. In most cases you do not need to install
+them yet; but sometime this may simplify the following steps.
+
+=item 3.
+
+Make sure that C<CPAN.pm> can download files from CPAN. If not, you may need
+to manually install C<Net::FTP>.
+
+=item 4.
+
+Install the bundle C<Bundle::OS2_default>
+
+ perl5.8.2 -MCPAN -e "install Bundle::OS2_default" < nul |& tee 00cpan_i_1
+
+This may take a couple of hours on 1GHz processor (when run the first time).
+And this should not be necessarily a smooth procedure. Some modules may not
+specify required dependencies, so one may need to repeat this procedure several
+times until the results stabilize.
+
+ perl5.8.2 -MCPAN -e "install Bundle::OS2_default" < nul |& tee 00cpan_i_2
+ perl5.8.2 -MCPAN -e "install Bundle::OS2_default" < nul |& tee 00cpan_i_3
+
+Even after they stabilize, some tests may fail.
+
+Fix as many discovered bugs as possible. Document all the bugs which are not
+fixed, and all the failures with unknown reasons. Inspect the produced logs
+F<00cpan_i_1> to find suspiciously skipped tests, and other fishy events.
+
+Keep in mind that I<installation> of some modules may fail too: for example,
+the DLLs to update may be already loaded by F<CPAN.pm>. Inspect the C<install>
+logs (in the example above F<00cpan_i_1> etc) for errors, and install things
+manually, as in
+
+ cd $CPANHOME/.cpan/build/Digest-MD5-2.31
+ make install
+
+Some distributions may fail some tests, but you may want to install them
+anyway (as above, or via C<force install> command of C<CPAN.pm> shell-mode).
+
+Since this procedure may take quite a long time to complete, it makes sense
+to "freeze" your CPAN configuration by disabling periodic updates of the
+local copy of CPAN index: set C<index_expire> to some big value (I use 365),
+then save the settings
+
+ CPAN> o conf index_expire 365
+ CPAN> o conf commit
+
+Reset back to the default value C<1> when you are finished.
+
+=item 5.
+
+When satisfied with the results, rerun the C<installcmd> target. Now you
+can copy C<perl5.8.2.exe> to C<perl.exe>, and install the other OMF-build
+executables: C<perl__.exe> etc. They are ready to be used.
+
+=item 6.
+
+Change to the C<./pod> directory of the build tree, download the Perl logo
+F<CamelGrayBig.BMP>, and run
+
+ ( perl2ipf > perl.ipf ) |& tee 00ipf
+ ipfc /INF perl.ipf |& tee 00inf
+
+This produces the Perl docs online book C<perl.INF>. Install in on
+C<BOOKSHELF> path.
+
+=item 7.
+
+Now is the time to build statically linked executable F<perl_.exe> which
+includes newly-installed via C<Bundle::OS2_default> modules. Doing testing
+via C<CPAN.pm> is going to be painfully slow, since it statically links
+a new executable per XS extension.
+
+Here is a possible workaround: create a toplevel F<Makefile.PL> in
+F<$CPANHOME/.cpan/build/> with contents being (compare with L<Making
+executables with a custom collection of statically loaded extensions>)
+
+ use ExtUtils::MakeMaker;
+ WriteMakefile NAME => 'dummy';
+
+execute this as
+
+ perl_5.8.2.exe Makefile.PL <nul |& tee 00aout_c1
+ make -k all test <nul |& 00aout_t1
+
+Again, this procedure should not be absolutely smooth. Some C<Makefile.PL>'s
+in subdirectories may be buggy, and would not run as "child" scripts. The
+interdependency of modules can strike you; however, since non-XS modules
+are already installed, the prerequisites of most modules have a very good
+chance to be present.
+
+If you discover some glitches, move directories of problematic modules to a
+different location; if these modules are non-XS modules, you may just ignore
+them - they are already installed; the remaining, XS, modules you need to
+install manually one by one.
+
+After each such removal you need to rerun the C<Makefile.PL>/C<make> process;
+usually this procedure converges soon. (But be sure to convert all the
+necessary external C libraries from F<.lib> format to F<.a> format: run one of
+
+ emxaout foo.lib
+ emximp -o foo.a foo.lib
+
+whichever is appropriate.) Also, make sure that the DLLs for external
+libraries are usable with with executables compiled without C<-Zmtd> options.
+
+When you are sure that only a few subdirectories
+lead to failures, you may want to add C<-j4> option to C<make> to speed up
+skipping subdirectories with already finished build.
+
+When you are satisfied with the results of tests, install the build C libraries
+for extensions:
+
+ make install |& tee 00aout_i
+
+Now you can rename the file F<./perl.exe> generated during the last phase
+to F<perl_5.8.2.exe>; place it on C<PATH>; if there is an inter-dependency
+between some XS modules, you may need to repeat the C<test>/C<install> loop
+with this new executable and some excluded modules - until the procedure
+converges.
+
+Now you have all the necessary F<.a> libraries for these Perl modules in the
+places where Perl builder can find it. Use the perl builder: change to an
+empty directory, create a "dummy" F<Makefile.PL> again, and run
+
+ perl_5.8.2.exe Makefile.PL |& tee 00c
+ make perl |& tee 00p
+
+This should create an executable F<./perl.exe> with all the statically loaded
+extensions built in. Compare the generated F<perlmain.c> files to make sure
+that during the iterations the number of loaded extensions only increases.
+Rename F<./perl.exe> to F<perl_5.8.2.exe> on C<PATH>.
+
+When it converges, you got a functional variant of F<perl_5.8.2.exe>; copy it
+to C<perl_.exe>. You are done with generation of the local Perl installation.
+
+=item 8.
+
+Make sure that the installed modules are actually installed in the location
+of the new Perl, and are not inherited from entries of @INC given for
+inheritance from the older versions of Perl: set C<PERLLIB_582_PREFIX> to
+redirect the new version of Perl to a new location, and copy the installed
+files to this new location. Redo the tests to make sure that the versions of
+modules inherited from older versions of Perl are not needed.
+
+Actually, the log output of L<pod2ipf> during the step 6 gives a very detailed
+info about which modules are loaded from which place; so you may use it as
+an additional verification tool.
+
+Check that some temporary files did not make into the perl install tree.
+Run something like this
+
+ pfind . -f "!(/\.(pm|pl|ix|al|h|a|lib|txt|pod|imp|bs|dll|ld|bs|inc|xbm|yml|cgi|uu|e2x|skip|packlist|eg|cfg|html|pub|enc|all|ini|po|pot)$/i or /^\w+$/") | less
+
+in the install tree (both top one and F<sitelib> one).
+
+Compress all the DLLs with F<lxlite>. The tiny F<.exe> can be compressed with
+C</c:max> (the bug only appears when there is a fixup in the last 6 bytes of a
+page (?); since the tiny executables are much smaller than a page, the bug
+will not hit). Do not compress C<perl_.exe> - it would not work under DOS.
+
+=item 9.
+
+Now you can generate the binary distribution. This is done by running the
+test of the CPAN distribution C<OS2::SoftInstaller>. Tune up the file
+F<test.pl> to suit the layout of current version of Perl first. Do not
+forget to pack the necessary external DLLs accordingly. Include the
+description of the bugs and test suite failures you could not fix. Include
+the small-stack versions of Perl executables from Perl build directory.
+
+Include F<perl5.def> so that people can relink the perl DLL preserving
+the binary compatibility, or can create compatibility DLLs. Include the diff
+files (C<diff -pu old new>) of fixes you did so that people can rebuild your
+version. Include F<perl5.map> so that one can use remote debugging.
+
+=item 10.
+
+Share what you did with the other people. Relax. Enjoy fruits of your work.
+
+=item 11.
+
+Brace yourself for thanks, bug reports, hate mail and spam coming as result
+of the previous step. No good deed should remain unpunished!
+
+=back
+
+=head1 Building custom F<.EXE> files
+
+The Perl executables can be easily rebuilt at any moment. Moreover, one can
+use the I<embedding> interface (see L<perlembed>) to make very customized
+executables.
+
+=head2 Making executables with a custom collection of statically loaded extensions
+
+It is a little bit easier to do so while I<decreasing> the list of statically
+loaded extensions. We discuss this case only here.
+
+=over
+
+=item 1.
+
+Change to an empty directory, and create a placeholder <Makefile.PL>:
+
+ use ExtUtils::MakeMaker;
+ WriteMakefile NAME => 'dummy';
+
+=item 2.
+
+Run it with the flavor of Perl (F<perl.exe> or F<perl_.exe>) you want to
+rebuild.
+
+ perl_ Makefile.PL
+
+=item 3.
+
+Ask it to create new Perl executable:
+
+ make perl
+
+(you may need to manually add C<PERLTYPE=-DPERL_CORE> to this commandline on
+some versions of Perl; the symptom is that the command-line globbing does not
+work from OS/2 shells with the newly-compiled executable; check with
+
+ .\perl.exe -wle "print for @ARGV" *
+
+).
+
+=item 4.
+
+The previous step created F<perlmain.c> which contains a list of newXS() calls
+near the end. Removing unnecessary calls, and rerunning
+
+ make perl
+
+will produce a customized executable.
+
+=back
+
+=head2 Making executables with a custom search-paths
+
+The default perl executable is flexible enough to support most usages.
+However, one may want something yet more flexible; for example, one may want
+to find Perl DLL relatively to the location of the EXE file; or one may want
+to ignore the environment when setting the Perl-library search patch, etc.
+
+If you fill comfortable with I<embedding> interface (see L<perlembed>), such
+things are easy to do repeating the steps outlined in L<Making
+executables with a custom collection of statically loaded extensions>, and
+doing more comprehensive edits to main() of F<perlmain.c>. The people with
+little desire to understand Perl can just rename main(), and do necessary
+modification in a custom main() which calls the renamed function in appropriate
+time.
+
+However, there is a third way: perl DLL exports the main() function and several
+callbacks to customize the search path. Below is a complete example of a
+"Perl loader" which
+
+=over
+
+=item 1.
+
+Looks for Perl DLL in the directory C<$exedir/../dll>;
+
+=item 2.
+
+Prepends the above directory to C<BEGINLIBPATH>;
+
+=item 3.
+
+Fails if the Perl DLL found via C<BEGINLIBPATH> is different from what was
+loaded on step 1; e.g., another process could have loaded it from C<LIBPATH>
+or from a different value of C<BEGINLIBPATH>. In these cases one needs to
+modify the setting of the system so that this other process either does not
+run, or loads the DLL from C<BEGINLIBPATH> with C<LIBPATHSTRICT=T> (available
+with kernels after September 2000).
+
+=item 4.
+
+Loads Perl library from C<$exedir/../dll/lib/>.
+
+=item 5.
+
+Uses Bourne shell from C<$exedir/../dll/sh/ksh.exe>.
+
+=back
+
+For best results compile the C file below with the same options as the Perl
+DLL. However, a lot of functionality will work even if the executable is not
+an EMX applications, e.g., if compiled with
+
+ gcc -Wall -DDOSISH -DOS2=1 -O2 -s -Zomf -Zsys perl-starter.c -DPERL_DLL_BASENAME=\"perl312F\" -Zstack 8192 -Zlinker /PM:VIO
+
+Here is the sample C file:
+
+ #define INCL_DOS
+ #define INCL_NOPM
+ /* These are needed for compile if os2.h includes os2tk.h, not os2emx.h */
+ #define INCL_DOSPROCESS
+ #include <os2.h>
+
+ #include "EXTERN.h"
+ #define PERL_IN_MINIPERLMAIN_C
+ #include "perl.h"
+
+ static char *me;
+ HMODULE handle;
+
+ static void
+ die_with(char *msg1, char *msg2, char *msg3, char *msg4)
+ {
+ ULONG c;
+ char *s = " error: ";
+
+ DosWrite(2, me, strlen(me), &c);
+ DosWrite(2, s, strlen(s), &c);
+ DosWrite(2, msg1, strlen(msg1), &c);
+ DosWrite(2, msg2, strlen(msg2), &c);
+ DosWrite(2, msg3, strlen(msg3), &c);
+ DosWrite(2, msg4, strlen(msg4), &c);
+ DosWrite(2, "\r\n", 2, &c);
+ exit(255);
+ }
+
+ typedef ULONG (*fill_extLibpath_t)(int type, char *pre, char *post, int replace, char *msg);
+ typedef int (*main_t)(int type, char *argv[], char *env[]);
+ typedef int (*handler_t)(void* data, int which);
+
+ #ifndef PERL_DLL_BASENAME
+ # define PERL_DLL_BASENAME "perl"
+ #endif
+
+ static HMODULE
+ load_perl_dll(char *basename)
+ {
+ char buf[300], fail[260];
+ STRLEN l, dirl;
+ fill_extLibpath_t f;
+ ULONG rc_fullname;
+ HMODULE handle, handle1;
+
+ if (_execname(buf, sizeof(buf) - 13) != 0)
+ die_with("Can't find full path: ", strerror(errno), "", "");
+ /* XXXX Fill `me' with new value */
+ l = strlen(buf);
+ while (l && buf[l-1] != '/' && buf[l-1] != '\\')
+ l--;
+ dirl = l - 1;
+ strcpy(buf + l, basename);
+ l += strlen(basename);
+ strcpy(buf + l, ".dll");
+ if ( (rc_fullname = DosLoadModule(fail, sizeof fail, buf, &handle)) != 0
+ && DosLoadModule(fail, sizeof fail, basename, &handle) != 0 )
+ die_with("Can't load DLL ", buf, "", "");
+ if (rc_fullname)
+ return handle; /* was loaded with short name; all is fine */
+ if (DosQueryProcAddr(handle, 0, "fill_extLibpath", (PFN*)&f))
+ die_with(buf, ": DLL exports no symbol ", "fill_extLibpath", "");
+ buf[dirl] = 0;
+ if (f(0 /*BEGINLIBPATH*/, buf /* prepend */, NULL /* append */,
+ 0 /* keep old value */, me))
+ die_with(me, ": prepending BEGINLIBPATH", "", "");
+ if (DosLoadModule(fail, sizeof fail, basename, &handle1) != 0)
+ die_with(me, ": finding perl DLL again via BEGINLIBPATH", "", "");
+ buf[dirl] = '\\';
+ if (handle1 != handle) {
+ if (DosQueryModuleName(handle1, sizeof(fail), fail))
+ strcpy(fail, "???");
+ die_with(buf, ":\n\tperl DLL via BEGINLIBPATH is different: \n\t",
+ fail,
+ "\n\tYou may need to manipulate global BEGINLIBPATH and LIBPATHSTRICT"
+ "\n\tso that the other copy is loaded via BEGINLIBPATH.");
+ }
+ return handle;
+ }
+
+ int
+ main(int argc, char **argv, char **env)
+ {
+ main_t f;
+ handler_t h;
+
+ me = argv[0];
+ /**/
+ handle = load_perl_dll(PERL_DLL_BASENAME);
+
+ if (DosQueryProcAddr(handle, 0, "Perl_OS2_handler_install", (PFN*)&h))
+ die_with(PERL_DLL_BASENAME, ": DLL exports no symbol ", "Perl_OS2_handler_install", "");
+ if ( !h((void *)"~installprefix", Perlos2_handler_perllib_from)
+ || !h((void *)"~dll", Perlos2_handler_perllib_to)
+ || !h((void *)"~dll/sh/ksh.exe", Perlos2_handler_perl_sh) )
+ die_with(PERL_DLL_BASENAME, ": Can't install @INC manglers", "", "");
+
+ if (DosQueryProcAddr(handle, 0, "dll_perlmain", (PFN*)&f))
+ die_with(PERL_DLL_BASENAME, ": DLL exports no symbol ", "dll_perlmain", "");
+ return f(argc, argv, env);
+ }
+
+
+=head1 Build FAQ
+
+=head2 Some C</> became C<\> in pdksh.
+
+You have a very old pdksh. See L<Prerequisites>.
+
+=head2 C<'errno'> - unresolved external
+
+You do not have MT-safe F<db.lib>. See L<Prerequisites>.
+
+=head2 Problems with tr or sed
+
+reported with very old version of tr and sed.
+
+=head2 Some problem (forget which ;-)
+
+You have an older version of F<perl.dll> on your LIBPATH, which
+broke the build of extensions.
+
+=head2 Library ... not found
+
+You did not run C<omflibs>. See L<Prerequisites>.
+
+=head2 Segfault in make
+
+You use an old version of GNU make. See L<Prerequisites>.
+
+=head2 op/sprintf test failure
+
+This can result from a bug in emx sprintf which was fixed in 0.9d fix 03.
+
+=head1 Specific (mis)features of OS/2 port
+
+=head2 C<setpriority>, C<getpriority>
+
+Note that these functions are compatible with *nix, not with the older
+ports of '94 - 95. The priorities are absolute, go from 32 to -95,
+lower is quicker. 0 is the default priority.
+
+B<WARNING>. Calling C<getpriority> on a non-existing process could lock
+the system before Warp3 fixpak22. Starting with Warp3, Perl will use
+a workaround: it aborts getpriority() if the process is not present.
+This is not possible on older versions C<2.*>, and has a race
+condition anyway.
+
+=head2 C<system()>
+
+Multi-argument form of C<system()> allows an additional numeric
+argument. The meaning of this argument is described in
+L<OS2::Process>.
+
+When finding a program to run, Perl first asks the OS to look for executables
+on C<PATH> (OS/2 adds extension F<.exe> if no extension is present).
+If not found, it looks for a script with possible extensions
+added in this order: no extension, F<.cmd>, F<.btm>,
+F<.bat>, F<.pl>. If found, Perl checks the start of the file for magic
+strings C<"#!"> and C<"extproc ">. If found, Perl uses the rest of the
+first line as the beginning of the command line to run this script. The
+only mangling done to the first line is extraction of arguments (currently
+up to 3), and ignoring of the path-part of the "interpreter" name if it can't
+be found using the full path.
+
+E.g., C<system 'foo', 'bar', 'baz'> may lead Perl to finding
+F<C:/emx/bin/foo.cmd> with the first line being
+
+ extproc /bin/bash -x -c
+
+If F</bin/bash.exe> is not found, then Perl looks for an executable F<bash.exe> on
+C<PATH>. If found in F<C:/emx.add/bin/bash.exe>, then the above system() is
+translated to
+
+ system qw(C:/emx.add/bin/bash.exe -x -c C:/emx/bin/foo.cmd bar baz)
+
+One additional translation is performed: instead of F</bin/sh> Perl uses
+the hardwired-or-customized shell (see C<L<"PERL_SH_DIR">>).
+
+The above search for "interpreter" is recursive: if F<bash> executable is not
+found, but F<bash.btm> is found, Perl will investigate its first line etc.
+The only hardwired limit on the recursion depth is implicit: there is a limit
+4 on the number of additional arguments inserted before the actual arguments
+given to system(). In particular, if no additional arguments are specified
+on the "magic" first lines, then the limit on the depth is 4.
+
+If Perl finds that the found executable is of PM type when the
+current session is not, it will start the new process in a separate session of
+necessary type. Call via C<OS2::Process> to disable this magic.
+
+B<WARNING>. Due to the described logic, you need to explicitly
+specify F<.com> extension if needed. Moreover, if the executable
+F<perl5.6.1> is requested, Perl will not look for F<perl5.6.1.exe>.
+[This may change in the future.]
+
+=head2 C<extproc> on the first line
+
+If the first chars of a Perl script are C<"extproc ">, this line is treated
+as C<#!>-line, thus all the switches on this line are processed (twice
+if script was started via cmd.exe). See L<perlrun/DESCRIPTION>.
+
+=head2 Additional modules:
+
+L<OS2::Process>, L<OS2::DLL>, L<OS2::REXX>, L<OS2::PrfDB>, L<OS2::ExtAttr>. These
+modules provide access to additional numeric argument for C<system>
+and to the information about the running process,
+to DLLs having functions with REXX signature and to the REXX runtime, to
+OS/2 databases in the F<.INI> format, and to Extended Attributes.
+
+Two additional extensions by Andreas Kaiser, C<OS2::UPM>, and
+C<OS2::FTP>, are included into C<ILYAZ> directory, mirrored on CPAN.
+Other OS/2-related extensions are available too.
+
+=head2 Prebuilt methods:
+
+=over 4
+
+=item C<File::Copy::syscopy>
+
+used by C<File::Copy::copy>, see L<File::Copy>.
+
+=item C<DynaLoader::mod2fname>
+
+used by C<DynaLoader> for DLL name mangling.
+
+=item C<Cwd::current_drive()>
+
+Self explanatory.
+
+=item C<Cwd::sys_chdir(name)>
+
+leaves drive as it is.
+
+=item C<Cwd::change_drive(name)>
+
+chanes the "current" drive.
+
+=item C<Cwd::sys_is_absolute(name)>
+
+means has drive letter and is_rooted.
+
+=item C<Cwd::sys_is_rooted(name)>
+
+means has leading C<[/\\]> (maybe after a drive-letter:).
+
+=item C<Cwd::sys_is_relative(name)>
+
+means changes with current dir.
+
+=item C<Cwd::sys_cwd(name)>
+
+Interface to cwd from EMX. Used by C<Cwd::cwd>.
+
+=item C<Cwd::sys_abspath(name, dir)>
+
+Really really odious function to implement. Returns absolute name of
+file which would have C<name> if CWD were C<dir>. C<Dir> defaults to the
+current dir.
+
+=item C<Cwd::extLibpath([type])>
+
+Get current value of extended library search path. If C<type> is
+present and positive, works with C<END_LIBPATH>, if negative, works
+with C<LIBPATHSTRICT>, otherwise with C<BEGIN_LIBPATH>.
+
+=item C<Cwd::extLibpath_set( path [, type ] )>
+
+Set current value of extended library search path. If C<type> is
+present and positive, works with <END_LIBPATH>, if negative, works
+with C<LIBPATHSTRICT>, otherwise with C<BEGIN_LIBPATH>.
+
+=item C<OS2::Error(do_harderror,do_exception)>
+
+Returns C<undef> if it was not called yet, otherwise bit 1 is
+set if on the previous call do_harderror was enabled, bit
+2 is set if on previous call do_exception was enabled.
+
+This function enables/disables error popups associated with
+hardware errors (Disk not ready etc.) and software exceptions.
+
+I know of no way to find out the state of popups I<before> the first call
+to this function.
+
+=item C<OS2::Errors2Drive(drive)>
+
+Returns C<undef> if it was not called yet, otherwise return false if errors
+were not requested to be written to a hard drive, or the drive letter if
+this was requested.
+
+This function may redirect error popups associated with hardware errors
+(Disk not ready etc.) and software exceptions to the file POPUPLOG.OS2 at
+the root directory of the specified drive. Overrides OS2::Error() specified
+by individual programs. Given argument undef will disable redirection.
+
+Has global effect, persists after the application exits.
+
+I know of no way to find out the state of redirection of popups to the disk
+I<before> the first call to this function.
+
+=item OS2::SysInfo()
+
+Returns a hash with system information. The keys of the hash are
+
+ MAX_PATH_LENGTH, MAX_TEXT_SESSIONS, MAX_PM_SESSIONS,
+ MAX_VDM_SESSIONS, BOOT_DRIVE, DYN_PRI_VARIATION,
+ MAX_WAIT, MIN_SLICE, MAX_SLICE, PAGE_SIZE,
+ VERSION_MAJOR, VERSION_MINOR, VERSION_REVISION,
+ MS_COUNT, TIME_LOW, TIME_HIGH, TOTPHYSMEM, TOTRESMEM,
+ TOTAVAILMEM, MAXPRMEM, MAXSHMEM, TIMER_INTERVAL,
+ MAX_COMP_LENGTH, FOREGROUND_FS_SESSION,
+ FOREGROUND_PROCESS
+
+=item OS2::BootDrive()
+
+Returns a letter without colon.
+
+=item C<OS2::MorphPM(serve)>, C<OS2::UnMorphPM(serve)>
+
+Transforms the current application into a PM application and back.
+The argument true means that a real message loop is going to be served.
+OS2::MorphPM() returns the PM message queue handle as an integer.
+
+See L<"Centralized management of resources"> for additional details.
+
+=item C<OS2::Serve_Messages(force)>
+
+Fake on-demand retrieval of outstanding PM messages. If C<force> is false,
+will not dispatch messages if a real message loop is known to
+be present. Returns number of messages retrieved.
+
+Dies with "QUITing..." if WM_QUIT message is obtained.
+
+=item C<OS2::Process_Messages(force [, cnt])>
+
+Retrieval of PM messages until window creation/destruction.
+If C<force> is false, will not dispatch messages if a real message loop
+is known to be present.
+
+Returns change in number of windows. If C<cnt> is given,
+it is incremented by the number of messages retrieved.
+
+Dies with "QUITing..." if WM_QUIT message is obtained.
+
+=item C<OS2::_control87(new,mask)>
+
+the same as L<_control87(3)> of EMX. Takes integers as arguments, returns
+the previous coprocessor control word as an integer. Only bits in C<new> which
+are present in C<mask> are changed in the control word.
+
+=item OS2::get_control87()
+
+gets the coprocessor control word as an integer.
+
+=item C<OS2::set_control87_em(new=MCW_EM,mask=MCW_EM)>
+
+The variant of OS2::_control87() with default values good for
+handling exception mask: if no C<mask>, uses exception mask part of C<new>
+only. If no C<new>, disables all the floating point exceptions.
+
+See L<"Misfeatures"> for details.
+
+=item C<OS2::DLLname([how [, \&xsub]])>
+
+Gives the information about the Perl DLL or the DLL containing the C
+function bound to by C<&xsub>. The meaning of C<how> is: default (2):
+full name; 0: handle; 1: module name.
+
+=back
+
+(Note that some of these may be moved to different libraries -
+eventually).
+
+
+=head2 Prebuilt variables:
+
+=over 4
+
+=item $OS2::emx_rev
+
+numeric value is the same as _emx_rev of EMX, a string value the same
+as _emx_vprt (similar to C<0.9c>).
+
+=item $OS2::emx_env
+
+same as _emx_env of EMX, a number similar to 0x8001.
+
+=item $OS2::os_ver
+
+a number C<OS_MAJOR + 0.001 * OS_MINOR>.
+
+=item $OS2::is_aout
+
+true if the Perl library was compiled in AOUT format.
+
+=item $OS2::can_fork
+
+true if the current executable is an AOUT EMX executable, so Perl can
+fork. Do not use this, use the portable check for
+$Config::Config{dfork}.
+
+=item $OS2::nsyserror
+
+This variable (default is 1) controls whether to enforce the contents
+of $^E to start with C<SYS0003>-like id. If set to 0, then the string
+value of $^E is what is available from the OS/2 message file. (Some
+messages in this file have an C<SYS0003>-like id prepended, some not.)
+
+=back
+
+=head2 Misfeatures
+
+=over 4
+
+=item *
+
+Since L<flock(3)> is present in EMX, but is not functional, it is
+emulated by perl. To disable the emulations, set environment variable
+C<USE_PERL_FLOCK=0>.
+
+=item *
+
+Here is the list of things which may be "broken" on
+EMX (from EMX docs):
+
+=over 4
+
+=item *
+
+The functions L<recvmsg(3)>, L<sendmsg(3)>, and L<socketpair(3)> are not
+implemented.
+
+=item *
+
+L<sock_init(3)> is not required and not implemented.
+
+=item *
+
+L<flock(3)> is not yet implemented (dummy function). (Perl has a workaround.)
+
+=item *
+
+L<kill(3)>: Special treatment of PID=0, PID=1 and PID=-1 is not implemented.
+
+=item *
+
+L<waitpid(3)>:
+
+ WUNTRACED
+ Not implemented.
+ waitpid() is not implemented for negative values of PID.
+
+=back
+
+Note that C<kill -9> does not work with the current version of EMX.
+
+=item *
+
+See L<"Text-mode filehandles">.
+
+=item *
+
+Unix-domain sockets on OS/2 live in a pseudo-file-system C</sockets/...>.
+To avoid a failure to create a socket with a name of a different form,
+C<"/socket/"> is prepended to the socket name (unless it starts with this
+already).
+
+This may lead to problems later in case the socket is accessed via the
+"usual" file-system calls using the "initial" name.
+
+=item *
+
+Apparently, IBM used a compiler (for some period of time around '95?) which
+changes FP mask right and left. This is not I<that> bad for IBM's
+programs, but the same compiler was used for DLLs which are used with
+general-purpose applications. When these DLLs are used, the state of
+floating-point flags in the application is not predictable.
+
+What is much worse, some DLLs change the floating point flags when in
+_DLLInitTerm() (e.g., F<TCP32IP>). This means that even if you do not I<call>
+any function in the DLL, just the act of loading this DLL will reset your
+flags. What is worse, the same compiler was used to compile some HOOK DLLs.
+Given that HOOK dlls are executed in the context of I<all> the applications
+in the system, this means a complete unpredictablity of floating point
+flags on systems using such HOOK DLLs. E.g., F<GAMESRVR.DLL> of B<DIVE>
+origin changes the floating point flags on each write to the TTY of a VIO
+(windowed text-mode) applications.
+
+Some other (not completely debugged) situations when FP flags change include
+some video drivers (?), and some operations related to creation of the windows.
+People who code B<OpenGL> may have more experience on this.
+
+Perl is generally used in the situation when all the floating-point
+exceptions are ignored, as is the default under EMX. If they are not ignored,
+some benign Perl programs would get a C<SIGFPE> and would die a horrible death.
+
+To circumvent this, Perl uses two hacks. They help against I<one> type of
+damage only: FP flags changed when loading a DLL.
+
+One of the hacks is to disable floating point exceptions on Perl startup (as
+is the default with EMX). This helps only with compile-time-linked DLLs
+changing the flags before main() had a chance to be called.
+
+The other hack is to restore FP flags after a call to dlopen(). This helps
+against similar damage done by DLLs _DLLInitTerm() at runtime. Currently
+no way to switch these hacks off is provided.
+
+=back
+
+=head2 Modifications
+
+Perl modifies some standard C library calls in the following ways:
+
+=over 9
+
+=item C<popen>
+
+C<my_popen> uses F<sh.exe> if shell is required, cf. L<"PERL_SH_DIR">.
+
+=item C<tmpnam>
+
+is created using C<TMP> or C<TEMP> environment variable, via
+C<tempnam>.
+
+=item C<tmpfile>
+
+If the current directory is not writable, file is created using modified
+C<tmpnam>, so there may be a race condition.
+
+=item C<ctermid>
+
+a dummy implementation.
+
+=item C<stat>
+
+C<os2_stat> special-cases F</dev/tty> and F</dev/con>.
+
+=item C<mkdir>, C<rmdir>
+
+these EMX functions do not work if the path contains a trailing C</>.
+Perl contains a workaround for this.
+
+=item C<flock>
+
+Since L<flock(3)> is present in EMX, but is not functional, it is
+emulated by perl. To disable the emulations, set environment variable
+C<USE_PERL_FLOCK=0>.
+
+=back
+
+=head2 Identifying DLLs
+
+All the DLLs built with the current versions of Perl have ID strings
+identifying the name of the extension, its version, and the version
+of Perl required for this DLL. Run C<bldlevel DLL-name> to find this
+info.
+
+=head2 Centralized management of resources
+
+Since to call certain OS/2 API one needs to have a correctly initialized
+C<Win> subsystem, OS/2-specific extensions may require getting C<HAB>s and
+C<HMQ>s. If an extension would do it on its own, another extension could
+fail to initialize.
+
+Perl provides a centralized management of these resources:
+
+=over
+
+=item C<HAB>
+
+To get the HAB, the extension should call C<hab = perl_hab_GET()> in C. After
+this call is performed, C<hab> may be accessed as C<Perl_hab>. There is
+no need to release the HAB after it is used.
+
+If by some reasons F<perl.h> cannot be included, use
+
+ extern int Perl_hab_GET(void);
+
+instead.
+
+=item C<HMQ>
+
+There are two cases:
+
+=over
+
+=item *
+
+the extension needs an C<HMQ> only because some API will not work otherwise.
+Use C<serve = 0> below.
+
+=item *
+
+the extension needs an C<HMQ> since it wants to engage in a PM event loop.
+Use C<serve = 1> below.
+
+=back
+
+To get an C<HMQ>, the extension should call C<hmq = perl_hmq_GET(serve)> in C.
+After this call is performed, C<hmq> may be accessed as C<Perl_hmq>.
+
+To signal to Perl that HMQ is not needed any more, call
+C<perl_hmq_UNSET(serve)>. Perl process will automatically morph/unmorph itself
+into/from a PM process if HMQ is needed/not-needed. Perl will automatically
+enable/disable C<WM_QUIT> message during shutdown if the message queue is
+served/not-served.
+
+B<NOTE>. If during a shutdown there is a message queue which did not disable
+WM_QUIT, and which did not process the received WM_QUIT message, the
+shutdown will be automatically cancelled. Do not call C<perl_hmq_GET(1)>
+unless you are going to process messages on an orderly basis.
+
+=item * Treating errors reported by OS/2 API
+
+There are two principal conventions (it is useful to call them C<Dos*>
+and C<Win*> - though this part of the function signature is not always
+determined by the name of the API) of reporting the error conditions
+of OS/2 API. Most of C<Dos*> APIs report the error code as the result
+of the call (so 0 means success, and there are many types of errors).
+Most of C<Win*> API report success/fail via the result being
+C<TRUE>/C<FALSE>; to find the reason for the failure one should call
+WinGetLastError() API.
+
+Some C<Win*> entry points also overload a "meaningful" return value
+with the error indicator; having a 0 return value indicates an error.
+Yet some other C<Win*> entry points overload things even more, and 0
+return value may mean a successful call returning a valid value 0, as
+well as an error condition; in the case of a 0 return value one should
+call WinGetLastError() API to distinguish a successful call from a
+failing one.
+
+By convention, all the calls to OS/2 API should indicate their
+failures by resetting $^E. All the Perl-accessible functions which
+call OS/2 API may be broken into two classes: some die()s when an API
+error is encountered, the other report the error via a false return
+value (of course, this does not concern Perl-accessible functions
+which I<expect> a failure of the OS/2 API call, having some workarounds
+coded).
+
+Obviously, in the situation of the last type of the signature of an OS/2
+API, it is must more convenient for the users if the failure is
+indicated by die()ing: one does not need to check $^E to know that
+something went wrong. If, however, this solution is not desirable by
+some reason, the code in question should reset $^E to 0 before making
+this OS/2 API call, so that the caller of this Perl-accessible
+function has a chance to distinguish a success-but-0-return value from
+a failure. (One may return undef as an alternative way of reporting
+an error.)
+
+The macros to simplify this type of error propagation are
+
+=over
+
+=item C<CheckOSError(expr)>
+
+Returns true on error, sets $^E. Expects expr() be a call of
+C<Dos*>-style API.
+
+=item C<CheckWinError(expr)>
+
+Returns true on error, sets $^E. Expects expr() be a call of
+C<Win*>-style API.
+
+=item C<SaveWinError(expr)>
+
+Returns C<expr>, sets $^E from WinGetLastError() if C<expr> is false.
+
+=item C<SaveCroakWinError(expr,die,name1,name2)>
+
+Returns C<expr>, sets $^E from WinGetLastError() if C<expr> is false,
+and die()s if C<die> and $^E are true. The message to die is the
+concatenated strings C<name1> and C<name2>, separated by C<": "> from
+the contents of $^E.
+
+=item C<WinError_2_Perl_rc>
+
+Sets C<Perl_rc> to the return value of WinGetLastError().
+
+=item C<FillWinError>
+
+Sets C<Perl_rc> to the return value of WinGetLastError(), and sets $^E
+to the corresponding value.
+
+=item C<FillOSError(rc)>
+
+Sets C<Perl_rc> to C<rc>, and sets $^E to the corresponding value.
+
+=back
+
+=item * Loading DLLs and ordinals in DLLs
+
+Some DLLs are only present in some versions of OS/2, or in some
+configurations of OS/2. Some exported entry points are present only
+in DLLs shipped with some versions of OS/2. If these DLLs and entry
+points were linked directly for a Perl executable/DLL or from a Perl
+extensions, this binary would work only with the specified
+versions/setups. Even if these entry points were not needed, the
+I<load> of the executable (or DLL) would fail.
+
+For example, many newer useful APIs are not present in OS/2 v2; many
+PM-related APIs require DLLs not available on floppy-boot setup.
+
+To make these calls fail I<only when the calls are executed>, one
+should call these API via a dynamic linking API. There is a subsystem
+in Perl to simplify such type of calls. A large number of entry
+points available for such linking is provided (see C<entries_ordinals>
+- and also C<PMWIN_entries> - in F<os2ish.h>). These ordinals can be
+accessed via the APIs:
+
+ CallORD(), DeclFuncByORD(), DeclVoidFuncByORD(),
+ DeclOSFuncByORD(), DeclWinFuncByORD(), AssignFuncPByORD(),
+ DeclWinFuncByORD_CACHE(), DeclWinFuncByORD_CACHE_survive(),
+ DeclWinFuncByORD_CACHE_resetError_survive(),
+ DeclWinFunc_CACHE(), DeclWinFunc_CACHE_resetError(),
+ DeclWinFunc_CACHE_survive(), DeclWinFunc_CACHE_resetError_survive()
+
+See the header files and the C code in the supplied OS/2-related
+modules for the details on usage of these functions.
+
+Some of these functions also combine dynaloading semantic with the
+error-propagation semantic discussed above.
+
+=back
+
+=head1 Perl flavors
+
+Because of idiosyncrasies of OS/2 one cannot have all the eggs in the
+same basket (though EMX environment tries hard to overcome this
+limitations, so the situation may somehow improve). There are 4
+executables for Perl provided by the distribution:
+
+=head2 F<perl.exe>
+
+The main workhorse. This is a chimera executable: it is compiled as an
+C<a.out>-style executable, but is linked with C<omf>-style dynamic
+library F<perl.dll>, and with dynamic CRT DLL. This executable is a
+VIO application.
+
+It can load perl dynamic extensions, and it can fork().
+
+B<Note.> Keep in mind that fork() is needed to open a pipe to yourself.
+
+=head2 F<perl_.exe>
+
+This is a statically linked C<a.out>-style executable. It cannot
+load dynamic Perl extensions. The executable supplied in binary
+distributions has a lot of extensions prebuilt, thus the above restriction is
+important only if you use custom-built extensions. This executable is a VIO
+application.
+
+I<This is the only executable with does not require OS/2.> The
+friends locked into C<M$> world would appreciate the fact that this
+executable runs under DOS, Win0.3*, Win0.95 and WinNT with an
+appropriate extender. See L<"Other OSes">.
+
+=head2 F<perl__.exe>
+
+This is the same executable as F<perl___.exe>, but it is a PM
+application.
+
+B<Note.> Usually (unless explicitly redirected during the startup)
+STDIN, STDERR, and STDOUT of a PM
+application are redirected to F<nul>. However, it is possible to I<see>
+them if you start C<perl__.exe> from a PM program which emulates a
+console window, like I<Shell mode> of Emacs or EPM. Thus it I<is
+possible> to use Perl debugger (see L<perldebug>) to debug your PM
+application (but beware of the message loop lockups - this will not
+work if you have a message queue to serve, unless you hook the serving
+into the getc() function of the debugger).
+
+Another way to see the output of a PM program is to run it as
+
+ pm_prog args 2>&1 | cat -
+
+with a shell I<different> from F<cmd.exe>, so that it does not create
+a link between a VIO session and the session of C<pm_porg>. (Such a link
+closes the VIO window.) E.g., this works with F<sh.exe> - or with Perl!
+
+ open P, 'pm_prog args 2>&1 |' or die;
+ print while <P>;
+
+The flavor F<perl__.exe> is required if you want to start your program without
+a VIO window present, but not C<detach>ed (run C<help detach> for more info).
+Very useful for extensions which use PM, like C<Perl/Tk> or C<OpenGL>.
+
+Note also that the differences between PM and VIO executables are only
+in the I<default> behaviour. One can start I<any> executable in
+I<any> kind of session by using the arguments C</fs>, C</pm> or
+C</win> switches of the command C<start> (of F<CMD.EXE> or a similar
+shell). Alternatively, one can use the numeric first argument of the
+C<system> Perl function (see L<OS2::Process>).
+
+=head2 F<perl___.exe>
+
+This is an C<omf>-style executable which is dynamically linked to
+F<perl.dll> and CRT DLL. I know no advantages of this executable
+over C<perl.exe>, but it cannot fork() at all. Well, one advantage is
+that the build process is not so convoluted as with C<perl.exe>.
+
+It is a VIO application.
+
+=head2 Why strange names?
+
+Since Perl processes the C<#!>-line (cf.
+L<perlrun/DESCRIPTION>, L<perlrun/Switches>,
+L<perldiag/"Not a perl script">,
+L<perldiag/"No Perl script found in input">), it should know when a
+program I<is a Perl>. There is some naming convention which allows
+Perl to distinguish correct lines from wrong ones. The above names are
+almost the only names allowed by this convention which do not contain
+digits (which have absolutely different semantics).
+
+=head2 Why dynamic linking?
+
+Well, having several executables dynamically linked to the same huge
+library has its advantages, but this would not substantiate the
+additional work to make it compile. The reason is the complicated-to-developers
+but very quick and convenient-to-users "hard" dynamic linking used by OS/2.
+
+There are two distinctive features of the dyna-linking model of OS/2:
+first, all the references to external functions are resolved at the compile time;
+second, there is no runtime fixup of the DLLs after they are loaded into memory.
+The first feature is an enormous advantage over other models: it avoids
+conflicts when several DLLs used by an application export entries with
+the same name. In such cases "other" models of dyna-linking just choose
+between these two entry points using some random criterion - with predictable
+disasters as results. But it is the second feature which requires the build
+of F<perl.dll>.
+
+The address tables of DLLs are patched only once, when they are
+loaded. The addresses of the entry points into DLLs are guaranteed to be
+the same for all the programs which use the same DLL. This removes the
+runtime fixup - once DLL is loaded, its code is read-only.
+
+While this allows some (significant?) performance advantages, this makes life
+much harder for developers, since the above scheme makes it impossible
+for a DLL to be "linked" to a symbol in the F<.EXE> file. Indeed, this
+would need a DLL to have different relocations tables for the
+(different) executables which use this DLL.
+
+However, a dynamically loaded Perl extension is forced to use some symbols
+from the perl
+executable, e.g., to know how to find the arguments to the functions:
+the arguments live on the perl
+internal evaluation stack. The solution is to put the main code of
+the interpreter into a DLL, and make the F<.EXE> file which just loads
+this DLL into memory and supplies command-arguments. The extension DLL
+cannot link to symbols in F<.EXE>, but it has no problem linking
+to symbols in the F<.DLL>.
+
+This I<greatly> increases the load time for the application (as well as
+complexity of the compilation). Since interpreter is in a DLL,
+the C RTL is basically forced to reside in a DLL as well (otherwise
+extensions would not be able to use CRT). There are some advantages if
+you use different flavors of perl, such as running F<perl.exe> and
+F<perl__.exe> simultaneously: they share the memory of F<perl.dll>.
+
+B<NOTE>. There is one additional effect which makes DLLs more wasteful:
+DLLs are loaded in the shared memory region, which is a scarse resource
+given the 512M barrier of the "standard" OS/2 virtual memory. The code of
+F<.EXE> files is also shared by all the processes which use the particular
+F<.EXE>, but they are "shared in the private address space of the process";
+this is possible because the address at which different sections
+of the F<.EXE> file are loaded is decided at compile-time, thus all the
+processes have these sections loaded at same addresses, and no fixup
+of internal links inside the F<.EXE> is needed.
+
+Since DLLs may be loaded at run time, to have the same mechanism for DLLs
+one needs to have the address range of I<any of the loaded> DLLs in the
+system to be available I<in all the processes> which did not load a particular
+DLL yet. This is why the DLLs are mapped to the shared memory region.
+
+=head2 Why chimera build?
+
+Current EMX environment does not allow DLLs compiled using Unixish
+C<a.out> format to export symbols for data (or at least some types of
+data). This forces C<omf>-style compile of F<perl.dll>.
+
+Current EMX environment does not allow F<.EXE> files compiled in
+C<omf> format to fork(). fork() is needed for exactly three Perl
+operations:
+
+=over 4
+
+=item *
+
+explicit fork() in the script,
+
+=item *
+
+C<open FH, "|-">
+
+=item *
+
+C<open FH, "-|">, in other words, opening pipes to itself.
+
+=back
+
+While these operations are not questions of life and death, they are
+needed for a lot of
+useful scripts. This forces C<a.out>-style compile of
+F<perl.exe>.
+
+
+=head1 ENVIRONMENT
+
+Here we list environment variables with are either OS/2- and DOS- and
+Win*-specific, or are more important under OS/2 than under other OSes.
+
+=head2 C<PERLLIB_PREFIX>
+
+Specific for EMX port. Should have the form
+
+ path1;path2
+
+or
+
+ path1 path2
+
+If the beginning of some prebuilt path matches F<path1>, it is
+substituted with F<path2>.
+
+Should be used if the perl library is moved from the default
+location in preference to C<PERL(5)LIB>, since this would not leave wrong
+entries in @INC. For example, if the compiled version of perl looks for @INC
+in F<f:/perllib/lib>, and you want to install the library in
+F<h:/opt/gnu>, do
+
+ set PERLLIB_PREFIX=f:/perllib/lib;h:/opt/gnu
+
+This will cause Perl with the prebuilt @INC of
+
+ f:/perllib/lib/5.00553/os2
+ f:/perllib/lib/5.00553
+ f:/perllib/lib/site_perl/5.00553/os2
+ f:/perllib/lib/site_perl/5.00553
+ .
+
+to use the following @INC:
+
+ h:/opt/gnu/5.00553/os2
+ h:/opt/gnu/5.00553
+ h:/opt/gnu/site_perl/5.00553/os2
+ h:/opt/gnu/site_perl/5.00553
+ .
+
+=head2 C<PERL_BADLANG>
+
+If 0, perl ignores setlocale() failing. May be useful with some
+strange I<locale>s.
+
+=head2 C<PERL_BADFREE>
+
+If 0, perl would not warn of in case of unwarranted free(). With older
+perls this might be
+useful in conjunction with the module DB_File, which was buggy when
+dynamically linked and OMF-built.
+
+Should not be set with newer Perls, since this may hide some I<real> problems.
+
+=head2 C<PERL_SH_DIR>
+
+Specific for EMX port. Gives the directory part of the location for
+F<sh.exe>.
+
+=head2 C<USE_PERL_FLOCK>
+
+Specific for EMX port. Since L<flock(3)> is present in EMX, but is not
+functional, it is emulated by perl. To disable the emulations, set
+environment variable C<USE_PERL_FLOCK=0>.
+
+=head2 C<TMP> or C<TEMP>
+
+Specific for EMX port. Used as storage place for temporary files.
+
+=head1 Evolution
+
+Here we list major changes which could make you by surprise.
+
+=head2 Text-mode filehandles
+
+Starting from version 5.8, Perl uses a builtin translation layer for
+text-mode files. This replaces the efficient well-tested EMX layer by
+some code which should be best characterized as a "quick hack".
+
+In addition to possible bugs and an inability to follow changes to the
+translation policy with off/on switches of TERMIO translation, this
+introduces a serious incompatible change: before sysread() on
+text-mode filehandles would go through the translation layer, now it
+would not.
+
+=head2 Priorities
+
+C<setpriority> and C<getpriority> are not compatible with earlier
+ports by Andreas Kaiser. See C<"setpriority, getpriority">.
+
+=head2 DLL name mangling: pre 5.6.2
+
+With the release 5.003_01 the dynamically loadable libraries
+should be rebuilt when a different version of Perl is compiled. In particular,
+DLLs (including F<perl.dll>) are now created with the names
+which contain a checksum, thus allowing workaround for OS/2 scheme of
+caching DLLs.
+
+It may be possible to code a simple workaround which would
+
+=over
+
+=item *
+
+find the old DLLs looking through the old @INC;
+
+=item *
+
+mangle the names according to the scheme of new perl and copy the DLLs to
+these names;
+
+=item *
+
+edit the internal C<LX> tables of DLL to reflect the change of the name
+(probably not needed for Perl extension DLLs, since the internally coded names
+are not used for "specific" DLLs, they used only for "global" DLLs).
+
+=item *
+
+edit the internal C<IMPORT> tables and change the name of the "old"
+F<perl????.dll> to the "new" F<perl????.dll>.
+
+=back
+
+=head2 DLL name mangling: 5.6.2 and beyond
+
+In fact mangling of I<extension> DLLs was done due to misunderstanding
+of the OS/2 dynaloading model. OS/2 (effectively) maintains two
+different tables of loaded DLL:
+
+=over
+
+=item Global DLLs
+
+those loaded by the base name from C<LIBPATH>; including those
+associated at link time;
+
+=item specific DLLs
+
+loaded by the full name.
+
+=back
+
+When resolving a request for a global DLL, the table of already-loaded
+specific DLLs is (effectively) ignored; moreover, specific DLLs are
+I<always> loaded from the prescribed path.
+
+There is/was a minor twist which makes this scheme fragile: what to do
+with DLLs loaded from
+
+=over
+
+=item C<BEGINLIBPATH> and C<ENDLIBPATH>
+
+(which depend on the process)
+
+=item F<.> from C<LIBPATH>
+
+which I<effectively> depends on the process (although C<LIBPATH> is the
+same for all the processes).
+
+=back
+
+Unless C<LIBPATHSTRICT> is set to C<T> (and the kernel is after
+2000/09/01), such DLLs are considered to be global. When loading a
+global DLL it is first looked in the table of already-loaded global
+DLLs. Because of this the fact that one executable loaded a DLL from
+C<BEGINLIBPATH> and C<ENDLIBPATH>, or F<.> from C<LIBPATH> may affect
+I<which> DLL is loaded when I<another> executable requests a DLL with
+the same name. I<This> is the reason for version-specific mangling of
+the DLL name for perl DLL.
+
+Since the Perl extension DLLs are always loaded with the full path,
+there is no need to mangle their names in a version-specific ways:
+their directory already reflects the corresponding version of perl,
+and @INC takes into account binary compatibility with older version.
+Starting from C<5.6.2> the name mangling scheme is fixed to be the
+same as for Perl 5.005_53 (same as in a popular binary release). Thus
+new Perls will be able to I<resolve the names> of old extension DLLs
+if @INC allows finding their directories.
+
+However, this still does not guarantee that these DLL may be loaded.
+The reason is the mangling of the name of the I<Perl DLL>. And since
+the extension DLLs link with the Perl DLL, extension DLLs for older
+versions would load an older Perl DLL, and would most probably
+segfault (since the data in this DLL is not properly initialized).
+
+There is a partial workaround (which can be made complete with newer
+OS/2 kernels): create a forwarder DLL with the same name as the DLL of
+the older version of Perl, which forwards the entry points to the
+newer Perl's DLL. Make this DLL accessible on (say) the C<BEGINLIBPATH> of
+the new Perl executable. When the new executable accesses old Perl's
+extension DLLs, they would request the old Perl's DLL by name, get the
+forwarder instead, so effectively will link with the currently running
+(new) Perl DLL.
+
+This may break in two ways:
+
+=over
+
+=item *
+
+Old perl executable is started when a new executable is running has
+loaded an extension compiled for the old executable (ouph!). In this
+case the old executable will get a forwarder DLL instead of the old
+perl DLL, so would link with the new perl DLL. While not directly
+fatal, it will behave the same as new executable. This beats the whole
+purpose of explicitly starting an old executable.
+
+=item *
+
+A new executable loads an extension compiled for the old executable
+when an old perl executable is running. In this case the extension
+will not pick up the forwarder - with fatal results.
+
+=back
+
+With support for C<LIBPATHSTRICT> this may be circumvented - unless
+one of DLLs is started from F<.> from C<LIBPATH> (I do not know
+whether C<LIBPATHSTRICT> affects this case).
+
+B<REMARK>. Unless newer kernels allow F<.> in C<BEGINLIBPATH> (older
+do not), this mess cannot be completely cleaned. (It turns out that
+as of the beginning of 2002, F<.> is not allowed, but F<.\.> is - and
+it has the same effect.)
+
+
+B<REMARK>. C<LIBPATHSTRICT>, C<BEGINLIBPATH> and C<ENDLIBPATH> are
+not environment variables, although F<cmd.exe> emulates them on C<SET
+...> lines. From Perl they may be accessed by L<Cwd::extLibpath> and
+L<Cwd::extLibpath_set>.
+
+=head2 DLL forwarder generation
+
+Assume that the old DLL is named F<perlE0AC.dll> (as is one for
+5.005_53), and the new version is 5.6.1. Create a file
+F<perl5shim.def-leader> with
+
+ LIBRARY 'perlE0AC' INITINSTANCE TERMINSTANCE
+ DESCRIPTION '@#perl5-porters@perl.org:5.006001#@ Perl module for 5.00553 -> Perl 5.6.1 forwarder'
+ CODE LOADONCALL
+ DATA LOADONCALL NONSHARED MULTIPLE
+ EXPORTS
+
+modifying the versions/names as needed. Run
+
+ perl -wnle "next if 0../EXPORTS/; print qq( \"$1\") if /\"(\w+)\"/" perl5.def >lst
+
+in the Perl build directory (to make the DLL smaller replace perl5.def
+with the definition file for the older version of Perl if present).
+
+ cat perl5shim.def-leader lst >perl5shim.def
+ gcc -Zomf -Zdll -o perlE0AC.dll perl5shim.def -s -llibperl
+
+(ignore multiple C<warning L4085>).
+
+=head2 Threading
+
+As of release 5.003_01 perl is linked to multithreaded C RTL
+DLL. If perl itself is not compiled multithread-enabled, so will not be perl's
+malloc(). However, extensions may use multiple thread on their own
+risk.
+
+This was needed to compile C<Perl/Tk> for XFree86-OS/2 out-of-the-box, and
+link with DLLs for other useful libraries, which typically are compiled
+with C<-Zmt -Zcrtdll>.
+
+=head2 Calls to external programs
+
+Due to a popular demand the perl external program calling has been
+changed wrt Andreas Kaiser's port. I<If> perl needs to call an
+external program I<via shell>, the F<f:/bin/sh.exe> will be called, or
+whatever is the override, see L<"PERL_SH_DIR">.
+
+Thus means that you need to get some copy of a F<sh.exe> as well (I
+use one from pdksh). The path F<F:/bin> above is set up automatically during
+the build to a correct value on the builder machine, but is
+overridable at runtime,
+
+B<Reasons:> a consensus on C<perl5-porters> was that perl should use
+one non-overridable shell per platform. The obvious choices for OS/2
+are F<cmd.exe> and F<sh.exe>. Having perl build itself would be impossible
+with F<cmd.exe> as a shell, thus I picked up C<sh.exe>. This assures almost
+100% compatibility with the scripts coming from *nix. As an added benefit
+this works as well under DOS if you use DOS-enabled port of pdksh
+(see L<"Prerequisites">).
+
+B<Disadvantages:> currently F<sh.exe> of pdksh calls external programs
+via fork()/exec(), and there is I<no> functioning exec() on
+OS/2. exec() is emulated by EMX by an asynchronous call while the caller
+waits for child completion (to pretend that the C<pid> did not change). This
+means that 1 I<extra> copy of F<sh.exe> is made active via fork()/exec(),
+which may lead to some resources taken from the system (even if we do
+not count extra work needed for fork()ing).
+
+Note that this a lesser issue now when we do not spawn F<sh.exe>
+unless needed (metachars found).
+
+One can always start F<cmd.exe> explicitly via
+
+ system 'cmd', '/c', 'mycmd', 'arg1', 'arg2', ...
+
+If you need to use F<cmd.exe>, and do not want to hand-edit thousands of your
+scripts, the long-term solution proposed on p5-p is to have a directive
+
+ use OS2::Cmd;
+
+which will override system(), exec(), C<``>, and
+C<open(,'...|')>. With current perl you may override only system(),
+readpipe() - the explicit version of C<``>, and maybe exec(). The code
+will substitute the one-argument call to system() by
+C<CORE::system('cmd.exe', '/c', shift)>.
+
+If you have some working code for C<OS2::Cmd>, please send it to me,
+I will include it into distribution. I have no need for such a module, so
+cannot test it.
+
+For the details of the current situation with calling external programs,
+see L<Starting OS/2 (and DOS) programs under Perl>. Set us mention a couple
+of features:
+
+=over 4
+
+=item *
+
+External scripts may be called by their basename. Perl will try the same
+extensions as when processing B<-S> command-line switch.
+
+=item *
+
+External scripts starting with C<#!> or C<extproc > will be executed directly,
+without calling the shell, by calling the program specified on the rest of
+the first line.
+
+=back
+
+=head2 Memory allocation
+
+Perl uses its own malloc() under OS/2 - interpreters are usually malloc-bound
+for speed, but perl is not, since its malloc is lightning-fast.
+Perl-memory-usage-tuned benchmarks show that Perl's malloc is 5 times quicker
+than EMX one. I do not have convincing data about memory footprint, but
+a (pretty random) benchmark showed that Perl's one is 5% better.
+
+Combination of perl's malloc() and rigid DLL name resolution creates
+a special problem with library functions which expect their return value to
+be free()d by system's free(). To facilitate extensions which need to call
+such functions, system memory-allocation functions are still available with
+the prefix C<emx_> added. (Currently only DLL perl has this, it should
+propagate to F<perl_.exe> shortly.)
+
+=head2 Threads
+
+One can build perl with thread support enabled by providing C<-D usethreads>
+option to F<Configure>. Currently OS/2 support of threads is very
+preliminary.
+
+Most notable problems:
+
+=over 4
+
+=item C<COND_WAIT>
+
+may have a race condition (but probably does not due to edge-triggered
+nature of OS/2 Event semaphores). (Needs a reimplementation (in terms of chaining
+waiting threads, with the linked list stored in per-thread structure?)?)
+
+=item F<os2.c>
+
+has a couple of static variables used in OS/2-specific functions. (Need to be
+moved to per-thread structure, or serialized?)
+
+=back
+
+Note that these problems should not discourage experimenting, since they
+have a low probability of affecting small programs.
+
+=head1 BUGS
+
+This description is not updated often (since 5.6.1?), see F<./os2/Changes>
+(L<perlos2delta>) for more info.
+
+=cut
+
+OS/2 extensions
+~~~~~~~~~~~~~~~
+I include 3 extensions by Andreas Kaiser, OS2::REXX, OS2::UPM, and OS2::FTP,
+into my ftp directory, mirrored on CPAN. I made
+some minor changes needed to compile them by standard tools. I cannot
+test UPM and FTP, so I will appreciate your feedback. Other extensions
+there are OS2::ExtAttr, OS2::PrfDB for tied access to EAs and .INI
+files - and maybe some other extensions at the time you read it.
+
+Note that OS2 perl defines 2 pseudo-extension functions
+OS2::Copy::copy and DynaLoader::mod2fname (many more now, see
+L<Prebuilt methods>).
+
+The -R switch of older perl is deprecated. If you need to call a REXX code
+which needs access to variables, include the call into a REXX compartment
+created by
+ REXX_call {...block...};
+
+Two new functions are supported by REXX code,
+ REXX_eval 'string';
+ REXX_eval_with 'string', REXX_function_name => \&perl_sub_reference;
+
+If you have some other extensions you want to share, send the code to
+me. At least two are available: tied access to EA's, and tied access
+to system databases.