From:	MERC::"uunet!CRVAX.SRI.COM!RELAY-INFO-VAX" 18-MAY-1992 20:50:14.87
To:	info-vax@kl.sri.com
CC:	
Subj:	LIB$FIND_FILE bug ? Help, please!


                          FNAL, MS #352, X-8591 (off.), 16-MAY-1992 - 17:32

    Hi, folks!

I detected a very strange bug (?) in LIB$FIND_FILE RTL routine. Please,
look at this simple test program for parsing wildcard files.

FORTRAN, sorry :-) .

C----------------------------------------------------------------------
      Program       Test
      Implicit      None
      Character*255 File /' '/
      Integer       LFile, IContext, Len, IChar
      Logical       IOK, Lib$Find_File
C
      Data IOK /.True./, IContext /0/, LFile /1/
C
    1 IOK = Lib$Find_File (
     &    '*.*.*',
     &    File,
     &    IContext,
     &    ,
     &    File(1:LFile),
     &    ,               )
      If ( IOK ) Then
        LFile = Len( File )
        Do While ( ( File(LFile:LFile)  .eq. ' ' ) .or. 
     &       ( IChar(File(LFile:LFile)) .eq.  0  ) )
          LFile = LFile - 1
        End Do
        Type '(1X,A)',File(1:LFile)
C        Open ( Unit = 1, File = File(1:LFile), Type = 'Old',
C     &         Readonly )
C
C ****    Here could be some stuff for working with the file contents
C
C        Close( Unit = 1 )
        Go To 1
      End If
      Call Lib$Find_File_End( IContext )
      Call Sys$GetMsg(%Val(IOK),LFile,File,,)
      Type '(1X,A)',File(1:LFile)
C
      End
C-----------------------------------------------------------------------

This program  works fine when the  Open - Close block  is commented. But if
you'd try to  uncomment it  the behaviour  became very  funny - the program
still works fine with the disk files. But whenever you try to run it on the
files on the TAPE which is mounted as a file device (i.e., without /FOREIGN
qualifier), the  program doesn't  exit after the all  files processing - it
goes into  infinite  loop,  rescanning  the tape  again and  again! Without
Open-Close, however, all is OK even with tapes. I use Exabyte tapes mounted
on the standard TK50 emulating drive. I didn't try it on ordinary tapes but
it seems that there is no difference.

I'm quite unhappy with  this funny bug - of  course it's possible to obtain
the list of all appropriate files first and then open them one-by-one using
this already  built list, but  that would  mean the  processing of the tape
twice, which is twice longer!

Now I'm playing with  the VAXCRTL routines and  may be they would help, but
still I don't understand the origin of this bug. I would be really grateful
to anybody who  helps with  this problem.  May be it is  real RTL bug and I
should send it to DEC?

I work under VMS V5.5. Sorry for previous empty posting

Thanx a lot in advance! Have fun!

Greg Landsberg
GLL@FNAL.FNAL.GOV
  ----------------------------------------------------------------------------
 | *******  Gregory L. Landsberg   ******* | E-mail (Internet):               |
 | (D0 Collaboration, GAMS Collaboration)  | Landsberg@sbhep2.phy.sunysb.edu  |
 | SUNY at Stony Brook, Physics dept.      | Landsberg@d0sb09.fnal.gov        |
 | Mail: MS 352 Fermilab Batavia IL 60510  | GLL@fnal.fnal.gov                |
 | FAX:   (708) 879-0048                   | E-mail (BITNET):                 |
 | Voice: (708) 840-3105 off, 840-4228 apt | Landsberg@SBHEPNY or GLL@FNAL    |
  ----------------------------------------------------------------------------