Fortran, debugger and Alpha/VMS 7.3-2 - VMS

This is a discussion on Fortran, debugger and Alpha/VMS 7.3-2 - VMS ; We finally upgraded the Alphas in our cluster from V7.3-1 to -2 last week. As expected, we never saw any problem until my colleague needed to use the debugger with her Fortran (V7.5...) program. It contains a Structure/record like structure ...

+ Reply to Thread
Results 1 to 16 of 16

Thread: Fortran, debugger and Alpha/VMS 7.3-2

  1. Fortran, debugger and Alpha/VMS 7.3-2

    We finally upgraded the Alphas in our cluster from V7.3-1 to -2 last
    week. As expected, we never saw any problem until my colleague needed
    to use the debugger with her Fortran (V7.5...) program.

    It contains a Structure/record like

    structure /asd$record/
    character*36 asd_name
    character*36 efile_name
    character*12 other_name
    ....
    end structure

    record /asd$record/ asd_record


    In the debugger we can

    EXA ASD_RECORD

    without problem but

    EXA ASD_RECORD.ASD_NAME

    generates the following error :-

    %DEBUG-E-INTERR, debugger error in DBGADDEXP\DETERMINE_TYPE unknown arg
    type or session corruption

    T'was fine on 7.3-1. Anybody know what's going on? John?

    Kristine, my colleague, is less than impressed.

    Cheers - Dave

    Now that I look at it, the $ in asd$record is a bit suspicious perhaps.

    --


  2. Re: Fortran, debugger and Alpha/VMS 7.3-2

    In article <6mnn33Fhnpk7U1@mid.individual.net>, "David Weatherall" writes:
    > We finally upgraded the Alphas in our cluster from V7.3-1 to -2 last
    > week. As expected, we never saw any problem until my colleague needed
    > to use the debugger with her Fortran (V7.5...) program.
    >
    > It contains a Structure/record like
    >
    > structure /asd$record/
    > character*36 asd_name
    > character*36 efile_name
    > character*12 other_name
    > ...
    > end structure
    >
    > record /asd$record/ asd_record


    I don't know what specifically is causing your problem, but I would
    not define anything with $ in the name. Generally $ should be
    considered reserved to VMS. You might be tripping over some VMS
    factility named asd (although I don't know of any).

    Try

    structure /asd_record_type/
    character*36 asd_name
    character*36 efile_name
    character*12 other_name
    ...
    end structure

    record /asd_record_type/ asd_record


  3. Re: Fortran, debugger and Alpha/VMS 7.3-2

    On Oct 28, 1:43*am, "David Weatherall" wrote:
    > We finally upgraded the Alphas in our cluster from V7.3-1 to -2 last
    > week. As expected, we never saw any problem until my colleague needed
    > to use the debugger with her Fortran (V7.5...) program.
    >
    > It contains a Structure/record like
    >
    > * * * * structure /asd$record/
    > * * * * * *character*36 * * * * asd_name
    > * * * * * *character*36 * * * * efile_name
    > * * * * * *character*12 * * * * other_name
    > ...
    > * * * * end structure
    >
    > * * * * record /asd$record/ asd_record
    >
    > In the debugger we can
    >
    > * * * * EXA ASD_RECORD
    >
    > without problem but
    >
    > * * * * EXA ASD_RECORD.ASD_NAME
    >
    > generates the following error :-
    >
    > %DEBUG-E-INTERR, debugger error in DBGADDEXP\DETERMINE_TYPE unknown arg
    > type or session corruption
    >
    > T'was fine on 7.3-1. Anybody know what's going on? John?
    >
    > Kristine, my colleague, is less than impressed.
    >
    > Cheers - Dave
    >
    > Now that I look at it, the $ in asd$record is a bit suspicious perhaps.
    >
    > --


    I just tried it on a 7.3-2 system and it worked.

    But, my Fortran ver is 7.2-180

    My fortran symbol was:

    "FORTRAN /NOLIST /DEBUG /NOOPT /NOALIGN /
    CHECK=(BOUN,FP_E,OVER,NOUNDE) /WARN=(NOALIGN,NOUNINIT)/EXTEND"


    Perhaps you should get your patches up to date if you have not done
    so. Mine are pretty close to up to date, (within a year or so
    maybe).




  4. Re: Fortran, debugger and Alpha/VMS 7.3-2

    On Oct 28, 1:43*am, "David Weatherall" wrote:
    > We finally upgraded the Alphas in our cluster from V7.3-1 to -2 last
    > week. As expected, we never saw any problem until my colleague needed
    > to use the debugger with her Fortran (V7.5...) program.
    >
    > It contains a Structure/record like
    >
    > * * * * structure /asd$record/
    > * * * * * *character*36 * * * * asd_name
    > * * * * * *character*36 * * * * efile_name
    > * * * * * *character*12 * * * * other_name
    > ...
    > * * * * end structure
    >
    > * * * * record /asd$record/ asd_record
    >
    > In the debugger we can
    >
    > * * * * EXA ASD_RECORD
    >
    > without problem but
    >
    > * * * * EXA ASD_RECORD.ASD_NAME
    >
    > generates the following error :-
    >
    > %DEBUG-E-INTERR, debugger error in DBGADDEXP\DETERMINE_TYPE unknown arg
    > type or session corruption
    >
    > T'was fine on 7.3-1. Anybody know what's going on? John?
    >
    > Kristine, my colleague, is less than impressed.
    >
    > Cheers - Dave
    >
    > Now that I look at it, the $ in asd$record is a bit suspicious perhaps.
    >
    > --


    I tried it on 7.3-2 with Compaq Fortran 77 V7.2-180-4293Q

    It worked fine. Complied with these switches:

    FOR*TRAN == "FORTRAN /NOLIST /DEBUG /NOOPT /NOALIGN /
    CHECK=(BOUN,FP_E,OVER,NOUNDE) /WARN=(NOALIGN,NOUNINIT)/EXTEND"

    Suggest you make sure your patches are up-to-date.

  5. Re: Fortran, debugger and Alpha/VMS 7.3-2

    Bob Koehler wrote:

    > In article <6mnn33Fhnpk7U1@mid.individual.net>, "David Weatherall"
    > writes:
    > > We finally upgraded the Alphas in our cluster from V7.3-1 to -2 last
    > > week. As expected, we never saw any problem until my colleague
    > > needed to use the debugger with her Fortran (V7.5...) program.
    > >
    > > It contains a Structure/record like
    > >
    > > structure /asd$record/
    > > character*36 asd_name
    > > character*36 efile_name
    > > character*12 other_name
    > > ...
    > > end structure
    > >
    > > record /asd$record/ asd_record

    >
    > I don't know what specifically is causing your problem, but I would
    > not define anything with $ in the name. Generally $ should be
    > considered reserved to VMS. You might be tripping over some VMS
    > factility named asd (although I don't know of any).
    >
    > Try
    >
    > structure /asd_record_type/
    > character*36 asd_name
    > character*36 efile_name
    > character*12 other_name
    > ...
    > end structure
    >
    > record /asd_record_type/ asd_record


    Thanks Bob
    had the same thought and it was the first thing I did when I
    got to work yesterday morning. Sadly no change. Our SysMan has raised a
    call on HP now. We'll see what they come up with.

    Cheers - Dave

    --


  6. Re: Fortran, debugger and Alpha/VMS 7.3-2

    tadamsmar wrote:

    > On Oct 28, 1:43*am, "David Weatherall" wrote:
    > > We finally upgraded the Alphas in our cluster from V7.3-1 to -2 last
    > > week. As expected, we never saw any problem until my colleague
    > > needed to use the debugger with her Fortran (V7.5...) program.
    > >
    > > It contains a Structure/record like
    > >
    > > * * * * structure /asd$record/
    > > * * * * * *character*36 * * * * asd_name
    > > * * * * * *character*36 * * * * efile_name
    > > * * * * * *character*12 * * * * other_name
    > > ...
    > > * * * * end structure
    > >
    > > * * * * record /asd$record/ asd_record
    > >
    > > In the debugger we can
    > >
    > > * * * * EXA ASD_RECORD
    > >
    > > without problem but
    > >
    > > * * * * EXA ASD_RECORD.ASD_NAME
    > >
    > > generates the following error :-
    > >
    > > %DEBUG-E-INTERR, debugger error in DBGADDEXP\DETERMINE_TYPE unknown
    > > arg type or session corruption
    > >
    > > T'was fine on 7.3-1. Anybody know what's going on? John?
    > >
    > > Kristine, my colleague, is less than impressed.
    > >
    > > Cheers - Dave
    > >
    > > Now that I look at it, the $ in asd$record is a bit suspicious
    > > perhaps.
    > >
    > > --

    >
    > I just tried it on a 7.3-2 system and it worked.
    >
    > But, my Fortran ver is 7.2-180
    >
    > My fortran symbol was:
    >
    > "FORTRAN /NOLIST /DEBUG /NOOPT /NOALIGN /
    > CHECK=(BOUN,FP_E,OVER,NOUNDE) /WARN=(NOALIGN,NOUNINIT)/EXTEND"
    >
    >
    > Perhaps you should get your patches up to date if you have not done
    > so. Mine are pretty close to up to date, (within a year or so
    > maybe).


    Thanks
    our Fortran version is 7.5 somethings. I've tried it with two
    standards. I didn't want to update our Fortran compiler to the latest
    standard because I'm in a position to do 'incremental Q/A' that way. If
    I upgrade and can't produce a bit-compatible .EXE I have to do the
    3-week Q/A job all over again. Partly because of the rules and partly
    for peace of mind.

    Thomas, our SysMAn, is sure that he applied all patches when he
    upgraded two weeks ago. Could you an ANALYZE /IMAGE on the debugger
    ..EXE please.

    Cheers - Dave.

    --


  7. Re: Fortran, debugger and Alpha/VMS 7.3-2

    On Oct 29, 1:17*am, "David Weatherall" wrote:
    > tadamsmar wrote:
    > > On Oct 28, 1:43*am, "David Weatherall" wrote:
    > > > We finally upgraded the Alphas in our cluster from V7.3-1 to -2 last
    > > > week. As expected, we never saw any problem until my colleague
    > > > needed to use the debugger with her Fortran (V7.5...) program.

    >
    > > > It contains a Structure/record like

    >
    > > > * * * * structure /asd$record/
    > > > * * * * * *character*36 * * * * asd_name
    > > > * * * * * *character*36 * * * * efile_name
    > > > * * * * * *character*12 * * * * other_name
    > > > ...
    > > > * * * * end structure

    >
    > > > * * * * record /asd$record/ asd_record

    >
    > > > In the debugger we can

    >
    > > > * * * * EXA ASD_RECORD

    >
    > > > without problem but

    >
    > > > * * * * EXA ASD_RECORD.ASD_NAME

    >
    > > > generates the following error :-

    >
    > > > %DEBUG-E-INTERR, debugger error in DBGADDEXP\DETERMINE_TYPE unknown
    > > > arg type or session corruption

    >
    > > > T'was fine on 7.3-1. Anybody know what's going on? John?

    >
    > > > Kristine, my colleague, is less than impressed.

    >
    > > > Cheers - Dave

    >
    > > > Now that I look at it, the $ in asd$record is a bit suspicious
    > > > perhaps.

    >
    > > > --

    >
    > > I just tried it on a 7.3-2 system and it worked.

    >
    > > But, my Fortran ver is 7.2-180

    >
    > > My fortran symbol was:

    >
    > > "FORTRAN /NOLIST /DEBUG /NOOPT /NOALIGN /
    > > CHECK=(BOUN,FP_E,OVER,NOUNDE) /WARN=(NOALIGN,NOUNINIT)/EXTEND"

    >
    > > Perhaps you should get your patches up to date if you have not done
    > > so. *Mine are pretty close to up to date, *(within a year or so
    > > maybe).

    >
    > Thanks
    > * * * *our Fortran version is 7.5 somethings. I've tried it with two
    > standards. I didn't want to update our Fortran compiler to the latest
    > standard because I'm in a position to do 'incremental Q/A' that way. If
    > I upgrade and can't produce a bit-compatible .EXE I have to do the
    > 3-week Q/A job all over again. Partly because of the rules and partly
    > for peace of mind.
    >
    > Thomas, our SysMAn, is sure that he applied all patches when he
    > upgraded two weeks ago. Could you an ANALYZE /IMAGE on the debugger
    > .EXE please.
    >
    > Cheers - Dave.
    >
    > --- Hide quoted text -
    >
    > - Show quoted text -


    Is the be part of the ANALYZE/IMAGE that you want?:

    Image Identification Information

    image name: "DEBUG"
    image file identification: "V7.3-200"
    image file build identification: "X9ZK-0060100000"
    link date/time: 1-OCT-2003 21:19:12.22
    linker identification: "A11-50"

    Patch Information

    There are no patches at this time.


  8. Re: Fortran, debugger and Alpha/VMS 7.3-2

    On Oct 29, 1:17*am, "David Weatherall" wrote:
    > tadamsmar wrote:
    > > On Oct 28, 1:43*am, "David Weatherall" wrote:
    > > > We finally upgraded the Alphas in our cluster from V7.3-1 to -2 last
    > > > week. As expected, we never saw any problem until my colleague
    > > > needed to use the debugger with her Fortran (V7.5...) program.

    >
    > > > It contains a Structure/record like

    >
    > > > * * * * structure /asd$record/
    > > > * * * * * *character*36 * * * * asd_name
    > > > * * * * * *character*36 * * * * efile_name
    > > > * * * * * *character*12 * * * * other_name
    > > > ...
    > > > * * * * end structure

    >
    > > > * * * * record /asd$record/ asd_record

    >
    > > > In the debugger we can

    >
    > > > * * * * EXA ASD_RECORD

    >
    > > > without problem but

    >
    > > > * * * * EXA ASD_RECORD.ASD_NAME

    >
    > > > generates the following error :-

    >
    > > > %DEBUG-E-INTERR, debugger error in DBGADDEXP\DETERMINE_TYPE unknown
    > > > arg type or session corruption

    >
    > > > T'was fine on 7.3-1. Anybody know what's going on? John?

    >
    > > > Kristine, my colleague, is less than impressed.

    >
    > > > Cheers - Dave

    >
    > > > Now that I look at it, the $ in asd$record is a bit suspicious
    > > > perhaps.

    >
    > > > --

    >
    > > I just tried it on a 7.3-2 system and it worked.

    >
    > > But, my Fortran ver is 7.2-180

    >
    > > My fortran symbol was:

    >
    > > "FORTRAN /NOLIST /DEBUG /NOOPT /NOALIGN /
    > > CHECK=(BOUN,FP_E,OVER,NOUNDE) /WARN=(NOALIGN,NOUNINIT)/EXTEND"

    >
    > > Perhaps you should get your patches up to date if you have not done
    > > so. *Mine are pretty close to up to date, *(within a year or so
    > > maybe).

    >
    > Thanks
    > * * * *our Fortran version is 7.5 somethings. I've tried it with two
    > standards. I didn't want to update our Fortran compiler to the latest
    > standard because I'm in a position to do 'incremental Q/A' that way. If
    > I upgrade and can't produce a bit-compatible .EXE I have to do the
    > 3-week Q/A job all over again. Partly because of the rules and partly
    > for peace of mind.
    >
    > Thomas, our SysMAn, is sure that he applied all patches when he
    > upgraded two weeks ago. Could you an ANALYZE /IMAGE on the debugger
    > .EXE please.
    >
    > Cheers - Dave.
    >
    > --- Hide quoted text -
    >
    > - Show quoted text -


    Here is the test program that I used, the code at the bottom was just
    left over from my previous t.for:

    structure /asd$record/
    character*36 asd_name
    character*36 efile_name
    character*12 other_name
    end structure


    record /asd$record/ asd_record



    type 1,-25500
    1 format (x,i6,'K')
    end

  9. Re: Fortran, debugger and Alpha/VMS 7.3-2

    tadamsmar wrote:

    > On Oct 29, 1:17*am, "David Weatherall" wrote:
    > > tadamsmar wrote:
    > > > On Oct 28, 1:43*am, "David Weatherall"
    > > > wrote:
    > > > > We finally upgraded the Alphas in our cluster from V7.3-1 to -2
    > > > > last week. As expected, we never saw any problem until my
    > > > > colleague needed to use the debugger with her Fortran (V7.5...)
    > > > > program.

    > >
    > > > > It contains a Structure/record like

    > >
    > > > > * * * * structure /asd$record/
    > > > > * * * * * *character*36 * * * * asd_name
    > > > > * * * * * *character*36 * * * * efile_name
    > > > > * * * * * *character*12 * * * * other_name
    > > > > ...
    > > > > * * * * end structure

    > >
    > > > > * * * * record /asd$record/ asd_record

    > >
    > > > > In the debugger we can

    > >
    > > > > * * * * EXA ASD_RECORD

    > >
    > > > > without problem but

    > >
    > > > > * * * * EXA ASD_RECORD.ASD_NAME

    > >
    > > > > generates the following error :-

    > >
    > > > > %DEBUG-E-INTERR, debugger error in DBGADDEXP\DETERMINE_TYPE
    > > > > unknown arg type or session corruption

    > >
    > > > > T'was fine on 7.3-1. Anybody know what's going on? John?

    > >
    > > > > Kristine, my colleague, is less than impressed.

    > >
    > > > > Cheers - Dave

    > >
    > > > > Now that I look at it, the $ in asd$record is a bit suspicious
    > > > > perhaps.

    > >
    > > > > --

    > >
    > > > I just tried it on a 7.3-2 system and it worked.

    > >
    > > > But, my Fortran ver is 7.2-180

    > >
    > > > My fortran symbol was:

    > >
    > > > "FORTRAN /NOLIST /DEBUG /NOOPT /NOALIGN /
    > > > CHECK=(BOUN,FP_E,OVER,NOUNDE) /WARN=(NOALIGN,NOUNINIT)/EXTEND"

    > >
    > > > Perhaps you should get your patches up to date if you have not
    > > > done so. *Mine are pretty close to up to date, *(within a year or
    > > > so maybe).

    > >
    > > Thanks
    > > * * * *our Fortran version is 7.5 somethings. I've tried it with two
    > > standards. I didn't want to update our Fortran compiler to the
    > > latest standard because I'm in a position to do 'incremental Q/A'
    > > that way. If I upgrade and can't produce a bit-compatible .EXE I
    > > have to do the 3-week Q/A job all over again. Partly because of the
    > > rules and partly for peace of mind.
    > >
    > > Thomas, our SysMAn, is sure that he applied all patches when he
    > > upgraded two weeks ago. Could you an ANALYZE /IMAGE on the debugger
    > > .EXE please.
    > >
    > > Cheers - Dave.
    > >
    > > --- Hide quoted text -
    > >
    > > - Show quoted text -

    >
    > Is the be part of the ANALYZE/IMAGE that you want?:
    >
    > Image Identification Information
    >
    > image name: "DEBUG"
    > image file identification: "V7.3-200"
    > image file build identification: "X9ZK-0060100000"
    > link date/time: 1-OCT-2003 21:19:12.22
    > linker identification: "A11-50"
    >
    > Patch Information
    >
    > There are no patches at this time.


    Thanks Mark
    that is exactly the bit I was after. My memory is that ours
    has link date in 2007. I'll forward this to work and check again when I
    get there.

    Cheers - Dave.

    --


  10. Re: Fortran, debugger and Alpha/VMS 7.3-2

    David Weatherall wrote:
    > tadamsmar wrote:
    >
    >
    >>On Oct 29, 1:17 am, "David Weatherall" wrote:
    >>
    >>>tadamsmar wrote:
    >>>
    >>>>On Oct 28, 1:43 am, "David Weatherall"
    >>>>wrote:
    >>>>
    >>>>>We finally upgraded the Alphas in our cluster from V7.3-1 to -2
    >>>>>last week. As expected, we never saw any problem until my
    >>>>>colleague needed to use the debugger with her Fortran (V7.5...)
    >>>>>program.
    >>>
    >>>>>It contains a Structure/record like
    >>>
    >>>>> structure /asd$record/
    >>>>> character*36 asd_name
    >>>>> character*36 efile_name
    >>>>> character*12 other_name
    >>>>>...
    >>>>> end structure
    >>>
    >>>>> record /asd$record/ asd_record
    >>>
    >>>>>In the debugger we can
    >>>
    >>>>> EXA ASD_RECORD
    >>>
    >>>>>without problem but
    >>>
    >>>>> EXA ASD_RECORD.ASD_NAME
    >>>
    >>>>>generates the following error :-
    >>>
    >>>>>%DEBUG-E-INTERR, debugger error in DBGADDEXP\DETERMINE_TYPE
    >>>>>unknown arg type or session corruption
    >>>
    >>>>>T'was fine on 7.3-1. Anybody know what's going on? John?
    >>>
    >>>>>Kristine, my colleague, is less than impressed.
    >>>
    >>>>>Cheers - Dave
    >>>
    >>>>>Now that I look at it, the $ in asd$record is a bit suspicious
    >>>>>perhaps.
    >>>
    >>>>>--
    >>>
    >>>>I just tried it on a 7.3-2 system and it worked.
    >>>
    >>>>But, my Fortran ver is 7.2-180
    >>>
    >>>>My fortran symbol was:
    >>>
    >>>>"FORTRAN /NOLIST /DEBUG /NOOPT /NOALIGN /
    >>>>CHECK=(BOUN,FP_E,OVER,NOUNDE) /WARN=(NOALIGN,NOUNINIT)/EXTEND"
    >>>
    >>>>Perhaps you should get your patches up to date if you have not
    >>>>done so. Mine are pretty close to up to date, (within a year or
    >>>>so maybe).
    >>>
    >>>Thanks
    >>> our Fortran version is 7.5 somethings. I've tried it with two
    >>>standards. I didn't want to update our Fortran compiler to the
    >>>latest standard because I'm in a position to do 'incremental Q/A'
    >>>that way. If I upgrade and can't produce a bit-compatible .EXE I
    >>>have to do the 3-week Q/A job all over again. Partly because of the
    >>>rules and partly for peace of mind.
    >>>
    >>>Thomas, our SysMAn, is sure that he applied all patches when he
    >>>upgraded two weeks ago. Could you an ANALYZE /IMAGE on the debugger
    >>>.EXE please.
    >>>
    >>>Cheers - Dave.
    >>>
    >>>--- Hide quoted text -
    >>>
    >>>- Show quoted text -

    >>
    >>Is the be part of the ANALYZE/IMAGE that you want?:
    >>
    >> Image Identification Information
    >>
    >> image name: "DEBUG"
    >> image file identification: "V7.3-200"
    >> image file build identification: "X9ZK-0060100000"
    >> link date/time: 1-OCT-2003 21:19:12.22
    >> linker identification: "A11-50"
    >>
    >> Patch Information
    >>
    >> There are no patches at this time.

    >
    >
    > Thanks Mark
    > that is exactly the bit I was after. My memory is that ours
    > has link date in 2007. I'll forward this to work and check again when I
    > get there.
    >
    > Cheers - Dave.
    >


    On my V7.3-2 system:

    Image Identification Information

    image name: "DEBUG"
    image file identification: "V8.3-008"
    image file build identification: "XA99-0060111007"
    link date/time: 14-JUN-2006 13:19:07.83
    linker identification: "A11-50"


    This came in the VMS732_DEBUG-V0100 ECO, which is included in the latest
    2 UPDATE ECOs, VMS732_UPDATE-V1500 and -V1600.

    Don't know if it fixes Mark's problem, though. One of the problem
    descriptions is:

    o An EXAMINE of a FORTRAN-90 deferred shape array would
    fail with a "descriptor not set up" error.

    Not exactly the same, but similar. Another is:

    An EXAMINE of a FORTRAN record or field within a record
    would lead to a debugger access violation.


    The third sounds exactly like the 1st, but with a typo in the
    description:

    An EXAMINE of a FORTRAN-90 deferred shape array would
    fail with a "descriptornot set up" error.

    HTH.

    --
    John Santos
    Evans Griffiths & Hart, Inc.
    781-861-0670 ext 539

  11. Re: Fortran, debugger and Alpha/VMS 7.3-2

    John Santos wrote:

    > David Weatherall wrote:
    > > tadamsmar wrote:
    > >
    > >
    > >>On Oct 29, 1:17 am, "David Weatherall"

    > wrote:
    > > >
    > > > > tadamsmar wrote:
    > > > >
    > > > > > On Oct 28, 1:43 am, "David Weatherall"
    > > > > > wrote:
    > > > > >
    > > > > > > We finally upgraded the Alphas in our cluster from V7.3-1
    > > > > > > to -2 last week. As expected, we never saw any problem
    > > > > > > until my colleague needed to use the debugger with her
    > > > > > > Fortran (V7.5...) program.
    > > > >
    > > > > > > It contains a Structure/record like
    > > > >
    > >>>>> structure /asd$record/
    > >>>>> character*36 asd_name
    > >>>>> character*36 efile_name
    > >>>>> character*12 other_name
    > > > > > > ...
    > >>>>> end structure
    > > > >
    > >>>>> record /asd$record/ asd_record
    > > > >
    > > > > > > In the debugger we can
    > > > >
    > >>>>> EXA ASD_RECORD
    > > > >
    > > > > > > without problem but
    > > > >
    > >>>>> EXA ASD_RECORD.ASD_NAME
    > > > >
    > > > > > > generates the following error :-
    > > > >
    > > > > > > %DEBUG-E-INTERR, debugger error in DBGADDEXP\DETERMINE_TYPE
    > > > > > > unknown arg type or session corruption
    > > > >
    > > > > > > T'was fine on 7.3-1. Anybody know what's going on? John?
    > > > >
    > > > > > > Kristine, my colleague, is less than impressed.
    > > > >
    > > > > > > Cheers - Dave
    > > > >
    > > > > > > Now that I look at it, the $ in asd$record is a bit
    > > > > > > suspicious perhaps.
    > > > >
    > > > > > > --
    > > > >
    > > > > > I just tried it on a 7.3-2 system and it worked.
    > > > >
    > > > > > But, my Fortran ver is 7.2-180
    > > > >
    > > > > > My fortran symbol was:
    > > > >
    > > > > > "FORTRAN /NOLIST /DEBUG /NOOPT /NOALIGN /
    > > > > > CHECK=(BOUN,FP_E,OVER,NOUNDE) /WARN=(NOALIGN,NOUNINIT)/EXTEND"
    > > > >
    > > > > > Perhaps you should get your patches up to date if you have not
    > > > > > done so. Mine are pretty close to up to date, (within a
    > > > > > year or so maybe).
    > > > >
    > > > > Thanks
    > >>> our Fortran version is 7.5 somethings. I've tried it with two
    > > > > standards. I didn't want to update our Fortran compiler to the
    > > > > latest standard because I'm in a position to do 'incremental
    > > > > Q/A' that way. If I upgrade and can't produce a bit-compatible
    > > > > .EXE I have to do the 3-week Q/A job all over again. Partly
    > > > > because of the rules and partly for peace of mind.
    > > > >
    > > > > Thomas, our SysMAn, is sure that he applied all patches when he
    > > > > upgraded two weeks ago. Could you an ANALYZE /IMAGE on the
    > > > > debugger .EXE please.
    > > > >
    > > > > Cheers - Dave.
    > > > >
    > > > > --- Hide quoted text -
    > > > >
    > > > > - Show quoted text -
    > > >
    > > > Is the be part of the ANALYZE/IMAGE that you want?:
    > > >
    > >> Image Identification Information
    > > >
    > >> image name: "DEBUG"
    > >> image file identification: "V7.3-200"
    > >> image file build identification: "X9ZK-0060100000"
    > >> link date/time: 1-OCT-2003 21:19:12.22
    > >> linker identification: "A11-50"
    > > >
    > >> Patch Information
    > > >
    > >> There are no patches at this time.

    > >
    > >
    > > Thanks Mark
    > > that is exactly the bit I was after. My memory is that
    > > ours has link date in 2007. I'll forward this to work and check
    > > again when I get there.
    > >
    > > Cheers - Dave.
    > >

    >
    > On my V7.3-2 system:
    >
    > Image Identification Information
    >
    > image name: "DEBUG"
    > image file identification: "V8.3-008"
    > image file build identification: "XA99-0060111007"
    > link date/time: 14-JUN-2006 13:19:07.83
    > linker identification: "A11-50"
    >
    >
    > This came in the VMS732_DEBUG-V0100 ECO, which is included in the
    > latest 2 UPDATE ECOs, VMS732_UPDATE-V1500 and -V1600.
    >
    > Don't know if it fixes Mark's problem, though. One of the problem
    > descriptions is:
    >
    > o An EXAMINE of a FORTRAN-90 deferred shape array
    > would fail with a "descriptor not set up" error.
    >
    > Not exactly the same, but similar. Another is:
    >
    > An EXAMINE of a FORTRAN record or field within a record
    > would lead to a debugger access violation.
    >
    >
    > The third sounds exactly like the 1st, but with a typo in the
    > description:
    >
    > An EXAMINE of a FORTRAN-90 deferred shape array would
    > fail with a "descriptornot set up" error.
    >
    > HTH.


    Thanks John
    I double checked our Image version yesterday. Ours is
    exactly the same as yours. I almost asked Thomas to undo it yesterday
    as I too started to curse. Fortunately, mine was just a little problem


    Will continue to wait a little while to see if HP come up with
    something but I have some Alpha changes to check out soon so I'll
    balance my colleague's frustration with my patience.

    We're still using the Fortran (not F90) compiler because of the VAX
    compatibility. I don't want F90 feature creeping into the code that
    would not then be available on VAX. It does make the code VMS-specific
    but all the RTL stuff does that anyway...

    Cheers - Dave.

    --


  12. Re: Fortran, debugger and Alpha/VMS 7.3-2

    David Weatherall wrote:
    >>>>>>> We finally upgraded the Alphas in our cluster from V7.3-1
    >>>>>>> to -2 last week. As expected, we never saw any problem
    >>>>>>> until my colleague needed to use the debugger with her
    >>>>>>> Fortran (V7.5...) program.
    >>>>>>> It contains a Structure/record like
    >>>>>>> structure /asd$record/
    >>>>>>> character*36 asd_name
    >>>>>>> character*36 efile_name
    >>>>>>> character*12 other_name
    >>>>>>> ...
    >>>>>>> end structure
    >>>>>>> record /asd$record/ asd_record
    >>>>>>> In the debugger we can
    >>>>>>> EXA ASD_RECORD
    >>>>>>> without problem but
    >>>>>>> EXA ASD_RECORD.ASD_NAME
    >>>>>>> generates the following error :-
    >>>>>>> %DEBUG-E-INTERR, debugger error in DBGADDEXP\DETERMINE_TYPE
    >>>>>>> unknown arg type or session corruption
    >>>>>>> T'was fine on 7.3-1. Anybody know what's going on? John?
    >>>>>>> Kristine, my colleague, is less than impressed.


    This is a bug and it was fixed by me many years ago. Pick up the recent
    patch kit for DEBUG and you should be fine.

    As John likes to say, here are the details "for those keeping score at
    home." The bug is not in the debugger, but in the Bliss compiler used to
    build DEBUG. The compiler was generating bad code for the routine
    DETERMINE_TYPE. I worked around the problem by disabling optimization.
    The workaround was removed for V8.3 because a newer compiler is used
    that fixes the bug.

    -Jeff Nelson

  13. Re: Fortran, debugger and Alpha/VMS 7.3-2

    In article , Jeff writes:
    >David Weatherall wrote:
    >>>>>>>> We finally upgraded the Alphas in our cluster from V7.3-1
    >>>>>>>> to -2 last week. As expected, we never saw any problem
    >>>>>>>> until my colleague needed to use the debugger with her
    >>>>>>>> Fortran (V7.5...) program.
    >>>>>>>> It contains a Structure/record like
    >>>>>>>> structure /asd$record/
    >>>>>>>> character*36 asd_name
    >>>>>>>> character*36 efile_name
    >>>>>>>> character*12 other_name
    >>>>>>>> ...
    >>>>>>>> end structure
    >>>>>>>> record /asd$record/ asd_record
    >>>>>>>> In the debugger we can
    >>>>>>>> EXA ASD_RECORD
    >>>>>>>> without problem but
    >>>>>>>> EXA ASD_RECORD.ASD_NAME
    >>>>>>>> generates the following error :-
    >>>>>>>> %DEBUG-E-INTERR, debugger error in DBGADDEXP\DETERMINE_TYPE
    >>>>>>>> unknown arg type or session corruption
    >>>>>>>> T'was fine on 7.3-1. Anybody know what's going on? John?
    >>>>>>>> Kristine, my colleague, is less than impressed.

    >
    >This is a bug and it was fixed by me many years ago. Pick up the recent
    >patch kit for DEBUG and you should be fine.
    >
    >As John likes to say, here are the details "for those keeping score at
    >home." The bug is not in the debugger, but in the Bliss compiler used to
    >build DEBUG. The compiler was generating bad code for the routine
    >DETERMINE_TYPE. I worked around the problem by disabling optimization.
    >The workaround was removed for V8.3 because a newer compiler is used
    >that fixes the bug.
    >
    >-Jeff Nelson


    Thanks Jeff....

    Any chance of getting that DELTA issue corrected on IA64 now that this
    issue has been put to bed?
    --
    VAXman- A Bored Certified VMS Kernel Mode Hacker VAXman(at)TMESIS(dot)COM

    .... pejorative statements of opinion are entitled to constitutional protection
    no matter how extreme, vituperous, or vigorously expressed they may be. (NJSC)

    Copr. 2008 Brian Schenkenberger. Publication of _this_ usenet article outside
    of usenet _must_ include its contents in its entirety including this copyright
    notice, disclaimer and quotations.

  14. Re: Fortran, debugger and Alpha/VMS 7.3-2

    VAXman- @SendSpamHere.ORG wrote:
    > Any chance of getting that DELTA issue corrected on IA64 now that this
    > issue has been put to bed?


    I'll see what I can do.

    -Jeff

  15. Re: Fortran, debugger and Alpha/VMS 7.3-2

    Jeff wrote:

    > David Weatherall wrote:
    > >>>>>>> We finally upgraded the Alphas in our cluster from V7.3-1
    > >>>>>>> to -2 last week. As expected, we never saw any problem
    > >>>>>>> until my colleague needed to use the debugger with her
    > >>>>>>> Fortran (V7.5...) program.
    > >>>>>>> It contains a Structure/record like
    > >>>>>>> structure /asd$record/
    > >>>>>>> character*36 asd_name
    > >>>>>>> character*36 efile_name
    > >>>>>>> character*12 other_name
    > >>>>>>> ...
    > >>>>>>> end structure
    > >>>>>>> record /asd$record/ asd_record
    > >>>>>>> In the debugger we can
    > >>>>>>> EXA ASD_RECORD
    > >>>>>>> without problem but
    > >>>>>>> EXA ASD_RECORD.ASD_NAME
    > >>>>>>> generates the following error :-
    > >>>>>>> %DEBUG-E-INTERR, debugger error in DBGADDEXP\DETERMINE_TYPE
    > >>>>>>> unknown arg type or session corruption
    > >>>>>>> T'was fine on 7.3-1. Anybody know what's going on? John?
    > >>>>>>> Kristine, my colleague, is less than impressed.

    >
    > This is a bug and it was fixed by me many years ago. Pick up the
    > recent patch kit for DEBUG and you should be fine.
    >
    > As John likes to say, here are the details "for those keeping score at
    > home." The bug is not in the debugger, but in the Bliss compiler used
    > to build DEBUG. The compiler was generating bad code for the routine
    > DETERMINE_TYPE. I worked around the problem by disabling optimization.
    > The workaround was removed for V8.3 because a newer compiler is used
    > that fixes the bug.
    >
    > -Jeff Nelson


    Thanks for the info Jeff. However, our SysMan tells me it is the latest
    patch. We've got this one loaded :-

    Image Identification Information

    image name: "DEBUG"
    image file identification: "V8.3-008"
    image file build identification: "XA99-0060111007"
    link date/time: 14-JUN-2006 13:19:07.83
    linker identification: "A11-50"

    Patch Information (C & P error - answer was None - DJW)

    Our initial answer from HP reflects your comment, ie, the F90 problem
    is fixed in the latest patch. However, I'm beginning to suspect that
    fixing the F90 problem has broken the F77 support and that's what we're
    still on. I'll give Thomas, our SysMan, the relevant F77/90 compiler
    versions for onward transmission on Monday. I started with the
    suspicion late Thursday night but I was tired and had too many hours in
    already. Friday I had off I did get as far as compiling the same
    Fortran module with both compilers, on the same machine, and ANALyzing
    the objects. There are some obvious differences in the GSD/DST records
    w.r.t Psect order etc. but I wouldn't expect myself to undertand much
    more than that. It just gave me an inkling of some of the differences,
    as I would expect the root of the problem to be the debugger's ability
    to process the Debug and Symbol information.

    btw - I presume you meant you disabled optimisation when using Bliss to
    compile DEBUG.

    Cheers - Dave.
    --

  16. Re: Fortran, debugger and Alpha/VMS 7.3-2

    David Weatherall wrote:

    > We finally upgraded the Alphas in our cluster from V7.3-1 to -2 last
    > week. As expected, we never saw any problem until my colleague needed
    > to use the debugger with her Fortran (V7.5...) program.
    >
    > It contains a Structure/record like
    >
    > structure /asd$record/
    > character*36 asd_name
    > character*36 efile_name
    > character*12 other_name
    > ...
    > end structure
    >
    > record /asd$record/ asd_record
    >
    >
    > In the debugger we can
    >
    > EXA ASD_RECORD
    >
    > without problem but
    >
    > EXA ASD_RECORD.ASD_NAME
    >
    > generates the following error :-
    >
    > %DEBUG-E-INTERR, debugger error in DBGADDEXP\DETERMINE_TYPE unknown
    > arg type or session corruption
    >
    > T'was fine on 7.3-1. Anybody know what's going on? John?
    >
    > Kristine, my colleague, is less than impressed.
    >
    > Cheers - Dave
    >
    > Now that I look at it, the $ in asd$record is a bit suspicious
    > perhaps.


    Just to let you all know where this went.

    Our system manager loaded a Patch over the w/e giving us the following
    version of DEBUG and DEBUGSHR :-

    Image Identification Information

    image name: "DEBUG"
    image file identification: "V8.3-015"
    image file build identification: "XBX2-SSB-0000"
    link date/time: 16-OCT-2008 08:16:56.94
    linker identification: "A11-50"

    Patch Information

    There are no patches at this time.


    It fixes the debugger's EXAMINE problem. My colleague is now 'ganz
    gluecklich und zufrieden' so thanks to whoever at HP resolved it
    (Jeff?).

    My only problem, now is that my command file that builds _my_
    application fails when Fortran-77 (V7.5) compiling my source with
    "/DEBUG". No /NOOPT. GEM whinges about GSD errors. I can't quote the
    precise error text 'cos I sent the mail with above information before
    finding the problem...

    I can't check what else the SM changed 'cos he's on a well-earned
    holiday and I've no desire to disturb him. I still debug on VAX so I
    can live with whatever side-effect of the DEBUG fix might be causing
    the compiler problem. I suspect compiling individual modules with get
    around it. ie..

    FOR /deb /sep mod1
    FOR /deb /sep mod2
    . . .
    FOR /deb /sep modn

    instead of

    FOR /deb /sep -
    !
    mod1,-
    mod2,-
    . . .
    modn

    as I've had similar problems in the past in another app,


    I Q/A my app next week so ...


    Cheers - Dave.

    --


+ Reply to Thread