1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.

Frame Issue When Compiling In Batch Mode

Discussion in 'Development' started by Razack, Feb 6, 2017.

  1. Razack

    Razack New Member

    When I compile a screen program in client session started by mpro. It works fine.

    Where as if I compile from _progres -b or mbpro session it gives a warning and doesn't create .r.

    ** Warning: Cannot fit FILL-IN Applicant with AT within FRAME F1. (4027)
    Warning: Ignoring position info for AT (2054)
    **BROWSE B1_APP will not fit in FRAME F1_APP in PROGRAM /home/XXX.p. (4028)


    Please advice. Also Why does progress has to check the frame and format when compiling. Shouldn't it be run time?
     
  2.  
  3. RealHeavyDude

    RealHeavyDude Well-Known Member

    Nope. Everything static - including frames - is resolved at compile time. Only dynamic handle based objects are resolved at runtime.
     
  4. Razack

    Razack New Member

    Agree that static objects will be resolved at compile time. From the perspective of Frames and display does it makes more sense to resolve at run time?

    What if I compile a program in a non batch mode and use that .r to run in a batch mode. Vice versa is my scenario .
     
  5. TomBascom

    TomBascom Curmudgeon

    It doesn't really matter if we think that FRAME and DISPLAY "ought" to be dynamic. As far as the language is concerned they are NOT.

    I must be getting old ;)

    You /can/ write code with dynamic frames. For instance, ProTop's lib/dynscreen.p lays out all of the screen widgets dynamically.
     
  6. RealHeavyDude

    RealHeavyDude Well-Known Member

    Don't get me wrong. It does not help to think about whether it makes sense or not. It has been that way since the beginning of the ABL in the 1980. But, as Tom already mentioned, beginning with V9 Progress has added dynamic handle-based objects to the language that you can create at runtime - hence these are resolved at runtime and the compiler does not bother with them.

    There are pros and cons to the static approach. One pro is that you don't need to manage the static objects - the AVM does that for you and that way you don't need to deal with potential memory leaks. Whereas, with dynamic handle-based objects it is up to you to manage them and you need to make sure that you are not generating memory leaks.

    For your requirement it might make sense to use dynamic frames instead.
     
  7. Razack

    Razack New Member

    Application that I work on is old and the thousands of programs already have the static frames. For the tool need of compiling the programs in the background it doesn't make sense to change the application programs.

    Is there are any other way in a batch session, we can compile the bunch programs using static frames ?

    Regards
     
  8. RealHeavyDude

    RealHeavyDude Well-Known Member

    Doing the compilation of UI programs in a background session ( -b ) means that there is no UI ( or to be precise the only UI is an output stream - usually a file ). The root of your issue is that the UI widget definitions are per desing not suited for a stream output. I don't understand why you insist to run the compile of a UI in batch mode. If you run it without the batch option you will be fine.

    Heavy Regards, RealHeavyDude.
     
  9. TomBascom

    TomBascom Curmudgeon

    Nightly batch compiles are a useful automation step.

    I'm not in a position to test right now but I faintly recall setting TERM=vt100 in the compile script as my solution to this problem.
     
  10. Razack

    Razack New Member

    Yes, I'm emphasizing on the batch compile for automation effort.
    Unfortunately I have not been able to find the solution yet. I had exported TERM to vt100 and still didn't resolve the issue.

    Regards
     
  11. I came across one such scenario today. It happens on unnamed frames; it didn't throw error until x(78). If I provide anything more that that compiling the program as batch process (_progres -b or mbpro) throws error whereas mpro doesn't. Why? Below is a simple 2 liner program that would demonstrate this issue.

    where compile.sh is;
    Code (progress):
    1. exec /opt/dlc/bin/_progres -b -p /home/sbalasub/compile/test.p
    where test1.p is;
    Code (progress):
    1. DEFINE VARIABLE file-name AS CHARACTER FORMAT "x(100)".
    2. UPDATE file-name.
    I see similar coding on multiple nightly jobs where they get input from file on the fly. Currently we compile using 'mpro', so no issues but if in case we had to automate we may be facing similar errors as well.
     
  12. TomBascom

    TomBascom Curmudgeon

    Nice simple test case! Thanks.

    That goaded my memory...

    Subject to correction from deep inside Progress I /think/ what happens is that during batch compiles Progress *assumes* an 80 character wide screen. Is your interactive "mpro" a wide screen? (That would explain why it works.)

    "100" blows up because it is more than 80 columns wide -- add "with width 102" to the update and the problem goes away (you need 2 spaces for the box drawn by default).

    (In real code use the actual width of the target device...)

    The TERM variable is apparently irrelevant. I must have been misremembering that.
     
    Last edited: Apr 6, 2017
  13. TomBascom

    TomBascom Curmudgeon

    Back to RealHeavyDude for a moment...

    IMHO we should be able to specify the *target* compile environment and not be restricted to only compiling for the running UI. I should, for instance, be able to compile UNIX batch r-code from a Windows GUI session. And vice-versa. It's really annoying and inconvenient to have to maintain a rat's nest of systems in order to support multiple platforms :(

    I'm not going to hold my breath waiting- but it sure would be nice.
     
  14. Razack

    Razack New Member

    Thanks Saravana for simple test case.

    @Tom Instead changing individual programs for compile purpose, is there a way to change the progress assumption of 80 width screen for a batch mode?

    Regards
     
  15. Rob Fitzpatrick

    Rob Fitzpatrick ProgressTalk.com Sponsor

  16. Razack

    Razack New Member

    Excellent Thanks Rob!! This is exactly I was looking for.

    Only drawback that we have is, we don't have the solution for progress version prior to 11.6.

    Regards
     
  17. TomBascom

    TomBascom Curmudgeon

    If you're not already running 11.6 then you're on an ancient, obsolete and unsupported release and you ought to upgrade... ;)
     
  18. Rob Fitzpatrick

    Rob Fitzpatrick ProgressTalk.com Sponsor

    Well, old releases are not technically unsupported... but you will definitely get better support the later a release you're on. And I agree with your other points. :)
     
  19. takeshi14121990

    takeshi14121990 New Member

    I think only dynamic handle based objects are resolved at runtime.
     

Share This Page