Conference tallis::alpha_migration_tools

    Why do you think revesting is necessary?


    I tried vesting your example, and I saw a lot of unsupported absolute
    references in the warnings.  We don't have FMS installed here so
    running it didn't do me a lot of good.  You may have an image which
    isn't vest-able.


    Brian

    
    what are all the sure signs to determine investability ? This
    list of negative signs is *VERY IMPORTANT* to determine whether
    vesting should be continued or not. 
    
    for instance, you indicate "unsupported absolute references", where
    to look for these errors ?
    
    This list will tell me not to waste time on further vesting but turning
    to recompilation as a 2nd alternative.
    
    when I vest, I got these warnings :

HANDVG::[] vest vaxsampcob.exe /warning
%VEST-I-NOHIF, HIF file SYS$COMMON:[SYSLIB]VAXSAMPCOB.HIF; not found
-RMS-E-FNF, file not found
%VEST-I-READING, Reading file SYS$COMMON:[SYSLIB]COBRTL.IIF;
%VEST-I-READING, Reading file SYS$COMMON:[SYSLIB]FDVSHR.IIF;
%VEST-I-READING, Reading file SYS$COMMON:[SYSLIB]LIBRTL2.IIF;
%VEST-I-READING, Reading file SYS$COMMON:[SYSLIB]LIBRTL.IIF;
%VEST-I-PASS1, Starting analysis pass 1
%VEST-I-VAXPACKED, VAX packed decimal string opcode CVTLP  -- will be emulated
        At:    SAMP\SAMP\1330 [COBOL] (0000A662)
        Input: CVTLP  1C34(R10),S^#08,(SP)
%VEST-I-VAXPACKED, VAX packed decimal string opcode EDITPC -- will be emulated
        At:    SAMP\SAMP\1330 [COBOL] (0000A668)
        Input: EDITPC S^#08,(R3),0FC4(R11),4884(R10)
%VEST-I-READING, Reading file SYS$COMMON:[SYSLIB]P1_SPACE.IIF;
%VEST-I-VAXPACKED, VAX packed decimal string opcode CVTTP  -- will be emulated
        At:    SAMP\SAMP\768 [COBOL] (0000968A)
        Input: CVTTP  S^#07,04(SP),@0000BB44,S^#06,0C(SP)
%VEST-I-FLAGIO, Image calls system service SYS$QIOW which may perform asynchrono
us I/O
        At:    COB$$RESTVA\_COB$CODE\263 [BLISS-32] (0000B691)
        Input: CALLS  S^#0C,@#7FFEDE00
%VEST-I-PASS2, Starting analysis pass 2
%VEST-I-ENDPASS2, Ending analysis pass 2 -- beginning code generation and output
%VEST-I-NONSTDCALLS, Non-standard call sets R10 R11
        At:    SAMP\SAMP\541 [COBOL] (00008E04)
%VEST-I-NONSTDCALLU, Non-standard call uses R10
        At:    PASSKY\PASSKY\2592 [COBOL] (0000AB6A)
%VEST-I-NONSTDCALLU, Non-standard call uses R10
        At:    RANGE\RANGE\3449 [COBOL] (0000AE73)
%VEST-I-NONSTDCALLU, Non-standard call uses R6
        At:    RANGE\RANGE\3478 [COBOL] (0000B057)

I have also placed the source file here where you can indicate 
abnormality :

HANDVC::SAMPCOB.COB;4

thanks.

-steve

    	First of all, you never answered my question as to why you think
    this image needs re-vesting.  Could you please elaborate?


    	If you get a lot of HIF entries, then you'll gain something by
    revesting.

    
>    what are all the sure signs to determine investability ? This
>    list of negative signs is *VERY IMPORTANT* to determine whether
>    vesting should be continued or not. 


    	I use the /SHOW=ALL/WARN=ALL qualifiers *always*.  Then, if you see
    informationals, check them over.  You're probably okay but you need to
    look at them carefully to make sure.  If you see warnings, there's a
    better chance you've got a problem.  And if you see errors, you've got
    problems for sure.


    	Please see Appendix C, Debugging Translations (in the translating
    guide), for more info.

    
>    for instance, you indicate "unsupported absolute references", where
>    to look for these errors ?


    	I may have gotten those because we don't have FMS installed here.
    I assume the image you vested in your last example was the same one as
    you had in HANDVC""::[], right?  If you translated with the qualifiers
    I used above, then you'll get *all* the warnings and so forth on your
    screen and in the .lis file vest will create.


    	For a list of the types of things vest doesn't support see the SPD.



    Brian

    
    >>First of all, you never answered my question as to why you think
    >>this image needs re-vesting.  Could you please elaborate?
    
    I got the impression that if you are not successful in your 1st pass
    of vesting an image, then you have to revest it a few more time, 
    this will increase the chance of vesting the image successfully.
    
    But then, I read the SPD as you indicated to me :
    
     " DECmigrate for OpenVMS AXP produces a working OpenVMS AXP executable
     from any OpenVMS VAX executable that meets the criteria outlined be-
     low. This translation is completed in a single pass, .........
                                           ^^^^^^^^^^^^^
    
     out accessing the program source code. Performance of the translated
                                            ^^^^^^^^^^^
     application can sometimes be improved by successive translations, us-
                                              ^^^^^^^^^^^^^^^^^^^^^^^ 
     ing the automatic feedback capability incorporated in DECmigrate. "
    
     The above description indicates translation is "complete" in the 1st
     pass and any subsequent passes only improve performance and not
     vestability. So if a image fails to translate on 1st pass, then no 
     need to continue vesting, right ? here vestability I define it as
     just making a VAX image runable on ALPHA regardless of performance. 
    
     My 2nd question then, I need an example to show me the techniques
     of improving a translated image performance via the proper use of 
     VEST qualifiers and interpretation of results. This part of 
     understanding VEST is not trivial, some common performance 
     enhancement examples are good aids to learning.  
       
     sl.
    
    
      

    
>    I got the impression that if you are not successful in your 1st pass
>    of vesting an image, then you have to revest it a few more time, 
>    this will increase the chance of vesting the image successfully.


    	No, you can only increase the performance, not the vestability.

        
>     The above description indicates translation is "complete" in the 1st
>     pass and any subsequent passes only improve performance and not
>     vestability. So if a image fails to translate on 1st pass, then no 
>     need to continue vesting, right ? here vestability I define it as
>     just making a VAX image runable on ALPHA regardless of performance. 


    	Yes.

    
>     My 2nd question then, I need an example to show me the techniques
>     of improving a translated image performance via the proper use of 
>     VEST qualifiers and interpretation of results. This part of 
>     understanding VEST is not trivial, some common performance 
>     enhancement examples are good aids to learning.  
       


    	If you get HIF entries written out while using the application,
    then retranslate.  This is described in section 4.2 of the book.



    Brian
          

    
    I have an image placed on handvc::GET_HWID.EXE;54
    
    I do the vesting and run the translated image and I got :
    
    [] run GET_HWID_TV.EXE;1
    status = 1
    hwid   = A-075-<8-
    len    = 9
    %TIE-I-HIFENTRIESWRITT, $1$DUA0:[USER0.][QA]GET_HWID.HIF; created. 
    There were 3 HIF entries written to it.
    
    can you take my program and show me with your session log on how
    to improve image performance via revesting ? maybe I didn't get 
    it right when I try it.
    
    sl.
    
     

No session log is necessary -- it's trivial to use a HIF file to improve performance.

Please read section 2.1.  There are various documented ways to make sure VEST sees your
HIF file when you retranslate.  Simplest is to move the HIF file to the directory where
you'll be VESTing, but you can also use qualifiers and logicals to point to the directory
where the HIF files are kept.  Once pointed to, you just reVEST.


Brian

You might also do a dir/title=hif in this notes conference.  This topic has been
discussed at length.


Brian