List:Internals« Previous MessageNext Message »
From:Michael Widenius Date:November 27 2000 10:11pm
Subject:Fwd: Re: Patches for OS/2 port 3.23.xx
View as plain text  
Hi!

Yuri> Hi,

Yuri> I forgot to add a description of my patches :-)

Yuri> getopt.h: added EMX definition of getopt.
Yuri> global.h: included os2.h at top: that's becase also OS2.H defines VOID
Yuri> (as void). But MySQL has a VOID(A). Compiling os2.h after MySQL
Yuri> definition will broke all VOID() uses.
Yuri> my_global.h: same as above.
Yuri> my_sys.h: added F_* definitions.
Yuri> mysql_com.h: now socket can defined at compile command line.
Yuri> thr_alarm.h: added proper typedef and correct macro/function
Yuri> definition.
Yuri> violite.c: ioctl() has 4 parameters also under OS/2.
Yuri> violite.c: missing r definition.
Yuri> ltconfig: added DLL support to libtool (untested)
Yuri> getopt.c: added EMX getopt() definition.
Yuri> mylock.c: flock undefined under emx.
Yuri> thr_alarm.c: changed OS2 macro
Yuri> 	- added missing function from unix code (untested)
Yuri> 	- changed return code check: rc==0 means NO_ERROR!
Yuri> 	- thr_got_alarm needs a pointer
Yuri> rltty.c: missing set_winsize.
Yuri> log_event.h: pthread.h write macro undef. Otherwise macro is applied to
Yuri> write() members in classes.
Yuri> mysqld.cc: reload_acl_and_cache() has new parameters.
Yuri> 	- fixed mistyping (= instead of ,)
Yuri> net_serv.cc: fixed thr_* calls.
Yuri> slave.cc: fixed thr_* calls.
Yuri> sql_class.h: pthread.h write macro undef. 
Yuri> sql_load.cc: ifdef around symlink.

Yuri> A few considerations about them:
Yuri> Major trouble is for thr_* functions: sometimes args are pointers to
Yuri> data, sometimes args are passed by value. I think that such functions
Yuri> should be properly rewritten, using the same system (e.g. arg by
Yuri> pointer) everywhere.

Thanks for the patch;  I have now applied this for 3.23.29-gamma.

I did also change the thr_alarm functions to take pointers; I also
changed thr_alarm.h to be a little more streamlined and I hope I
didn't break anything for OS/2 when I did this ;  I would appreciate
if you could check out the code in 3.23.29 if I got this right!

Regards,
Monty
Thread
Fwd: Re: Patches for OS/2 port 3.23.xxMichael Widenius27 Nov
  • Re: Patches for OS/2 port 3.23.xxYuri Dario28 Nov
    • Re: Patches for OS/2 port 3.23.xxSasha Pachev28 Nov
      • Re: Patches for OS/2 port 3.23.xxYuri Dario29 Nov