Beta 2 on HPUX and other problems ...

[Ken Stone]

On HPUX 10.20, with the beta2 bits, PRINTCAP_NAME never gets defined so
the compile bombs with ...

[ROOT @ z] (25)-> make
Using FLAGS = -Ae -Iinclude -I./include -I./ubiqx -I./smbwrapper -Ae -DSMBLOGFIL
E="/local/var/log.smb" -DNMBLOGFILE="/local/var/log.nmb" -DCONFIGFILE="/local/li
b/smb.conf" -DLMHOSTSFILE="/local/lib/lmhosts"   -DSWATDIR="/local/swat" -DSBIND
IR="/local/bin" -DLOCKDIR="/local/var/locks" -DSMBRUN="/local/bin/smbrun" -DCODE
PAGEDIR="/local/lib/codepages" -DDRIVERFILE="/local/lib/printers.def" -DBINDIR="
/local/bin" -DHAVE_INCLUDES_H -DPASSWD_PROGRAM="/bin/passwd" -DSMB_PASSWD_FILE="
/local/private/smbpasswd" -DSMB_PASSWD_PROGRAM="/local/bin/smbpasswd"
Using LIBS = -lsec
Compiling param/loadparm.c
cc: "param/loadparm.c", line 820: error 1588: "PRINTCAP_NAME" undefined.
cc: "param/loadparm.c", line 820: warning 563: Argument #2 is not the correct ty
pe.
*** Error exit code 1

Stop.

Looks like the issue is in how it is handled in include/includes.h

First ....

#ifdef AIX
#define DEFAULT_PRINTING PRINT_AIX
#define PRINTCAP_NAME "/etc/qconfig"
#endif

#ifdef HPUX
#define DEFAULT_PRINTING PRINT_HPUX
#endif

Notice how it is defined for AIX but not HPUX and then farther down

#ifndef DEFAULT_PRINTING
#ifdef SYSV
#define DEFAULT_PRINTING PRINT_SYSV
#define PRINTCAP_NAME "lpstat"
#else
#define DEFAULT_PRINTING PRINT_BSD
#define PRINTCAP_NAME "/etc/printcap"
#endif
#endif

Looks like it just got dropped somewhere.

Only places its found are 

[ROOT @ z] (26)-> find . -name '*.[ch]' | xargs grep PRINTCAP_NAME
./param/loadparm.c:  string_set(&Globals.szPrintcapname, PRINTCAP_NAME);
./include/includes.h:#define PRINTCAP_NAME "/etc/qconfig"
./include/includes.h:#define PRINTCAP_NAME "lpstat"
./include/includes.h:#define PRINTCAP_NAME "/etc/printcap"
./utils/testprns.c:         pszTemp = (argc < 3) ? PRINTCAP_NAME : argv[2];
[ROOT @ z] (27)->



And another problem I'm having on any/all of the 2.X stuff.  This is only
on win95, not NT, not win98.

Scenario is Win95 client opens a file on samba server (our test case is a .txt 
file with notepad).  Win95 client closes file (no mods to file needed).  
Checking status on swat after Win95 client closes file show that file is still 
locked.  Now modify the .txt file on the samba server (ie vi it and make a 
change).  Now on the Win95 client using explorer, you can refresh the samba 
server directory with F5 and see the mtime on the file has changed ... but ... 
if you edit the file again on the Win95 client with notepad, you get the old 
file, not the changed version.  Odd thing here is that if you do one of the 
following

    * disconnect/reconnect the samba share
    * go to an NT or win98 client and edit the file with notepad (just open
      and close it, no mods)
    * kill the smbd running for the Win95 client (then let the client auto
      re-connect)

and then try and edit the file on the Win95 client ... you get the right 
version, ie, the one modified on the UNIX side ?

This is driving us nuts ...

Thanks

  -- Ken