Re: IND$FILE Update

--- In [email protected], "Rocky" <rocsystems@...> wrote:
>
> Hi all
>  
> The GCIC  full screen  emulator is now working with a "newer version" of the
> IND$FILE emulation.
>  
> I have tested it with VISTA and WC3270 (x3270) and both work fine for
> uploading.
> Please do some testing and let me know the results.
>  
> Before I continue - Now is the time for some KUDO's to Tom Brennon - The
> Vista Developer.
>  
> I ran into some serious problems in the logic and when I got to wits end, I
> wrote Tom directly and he very genrously gave me some pointers about the
> logic.
>  
> We don't see this kind of help too often any more  outside of the Hercules
> forums and family. 
> Especially since his application is commercial.
>  
> He invested some time - dug out his old code and spent part if his weekend
> putting some stuff together for me. 
> I very much appreciated his efforts, and since I never had any contact with
> him before, its worth mentioning his genrous help.
>  
> Aside from the above - (I have been using a different Emulator for years but
> I was very impressed with the VISTA - very nice product) 
>  
> Back to the IND$FILE issues:
> The Upload is working with some caveats - It will default to LRECL 130
> variable unless you enter the parameters.
> Want some feedback before I start the download -- No since writing this
> twice if we have problems.
>  
> It seems that that there are two different base translation tables that work
> with IND$FILE.
>  
> The original table that we found on the open source C3270 site - (FT_CUT.C)
> 
> I used that LOGIC first and it worked with my very old (25 years) LOGNET
> EMULATOR both on the real IND$FILE and on the GCIC emulator in VM/370.
> However, newer emulators Like VISTA (less than 25 years) or X3270 and
> probably others, did not work on the GCIC application but did work agains
> the real IND$FILE   
> 
>  
> I examined the traces from these emulators and I could see that they were
> using different translation tables.
> Enter Tom Brennon ---
>  
> Using Tom's information, I changed some of the conversion logic and Now the
> GCIC is working with VISTA and WC3270 but not with my old emulator.
>  
> There are probably a few other emulators out there in the community, like
> PC3270 or ATTACHMATE etc. Like to know if they are working properly so
> please give it a try.
>  
> Best I can figure out, is that somewhere along the line, IND$FILE logic was
> changed - The newer emulators are using the newer logic, but the older
> emulators are still using the old logic and the real IND$FILE was made
> backward compatible to handle both. Thats just a guess
>  
> A few questions arise with this logic:
>  
> 1) I can't tell from the traces how the Real IND$FILE knows which tables to
> use. They are incompatible.
> 2) The logic that I saw in the X3270 source does not match what X3270 is
> sending. However it did match what my old LOGNET emulation is sending.
> (Source code management)
>  
> All very strange - I'll do some more investigating - But if the ony problem
> is my old emulator - Either I'll switch to Vista- or add the old routine to
> the application and instead of using IND$FILE I'll use IND&FILE to use the
> oldroutine if I cant find a better way.
>  
> Server is up with the new code.   rocsys.dyndns.org port 3277  D  CMSUSER
> pw TSTUSR1 TSTUSR1
> File transfer is accessed from FLIST screen  - PF6
> Tab character transfers as x'05'   - the CH /x'05/    / * * works - just as
> written (1 apostrophe) - for the "C" programmers
>  
> When you get back to FLIST after the transfer - if the file  is new you will
> have to do a refresh. 
> SD will show the files in descending date order
> "A" disk is R/W
>  
> If you fill it up please remove most of the data when you are done. 
>  
> Questions - always
>  
> Once again many thx to Tom
>  
> Roc
>  
Hi Roc,

here are the results of some tests (with nice Brennan's VISTA):

1) Transferred an ASSEMBLER source (ASCII) to GIND: OK.

2) Transferred an ASM 386 source to GIND: OK (some mess with TABS..;-)

3) Transferred an ASSEMBLER source in EBCDIC to GIND: file transfer OK but I couldn't find it in FLIST; after some investigation I found it had overwritten the last file transferred; anyhow with unintellegible data. 

4) Transferred again the same EBCDIC source: now I could find it but unintellegible.

Transferred again the ASM 386 source: OK (but for TABS).

By all means great improvement; go ahead Roc!

Paulo