Conference iosg::all-in-1_v30

    Hans,
    
    Did they rerun pre-check after resetting the parameters and of course
    reboot if the paramters were non-dynamic.
    
    pre-check is really worth the time and effort it takes.
    
    regards,
    
    Andrew.D.Wicks

    Hallo,
    
    I don't think INSVIRMEM produced by the LINKer is
    caused by (process) quota.
    
    Are there third party integrations ?
    
    I has to do with user-written based images. Don't ask me for
    the details. See VMS note files:
    
SYSAUX:[NOTES$LIBRARY]VMSNOTES_V1.DRS;5

1705-  2-851212- LINK-F-INSVIRMEM is different!

SYSAUX:[NOTES$LIBRARY]VMSNOTES_V3.DRS;3

 106-  1-870224- LINKER and INSVIRMEM

SYSAUX:[NOTES$LIBRARY]VMSNOTES_V6.DRS;2

 244-  3-890123- LINK-E-INSVIRMEM
    
    Regards,
    
    Jan
    

    Well, parameters and quotas were OK, so .3 seems to be the answer.
    
    I talked to our VMS group, and they suggested linking with 
    LINK/FULL/MAP to find the problem. 
    
    Question is can you do this? I guess you'd have mess around with
    A1LINK.COM and change the LINK commands in there? But since linking
    is included in installation of 3.0.... Can you install without 
    linking?
    
    Hans

    If you're linking outside the image, there are options you can pass to
    A1LINK (well actually A1LNKDRV) to cause link maps to be generated.
    
    Since this link is inside the installation, and there's no way that you
    can tell the installation to *NOT* do the link, or carry on if it
    fails, you are going to have to do some hacking of the kit:
    
    	1) Change A1LINK (In A1LUS030.D) to make it look like the map has
    	been asked for. You'll see it checking some symbols at the top.
    
    	2) Change KIIMGAES.COM (In A1030.A) to turn off error checking
    	around the link.
    
    Change 1 won't stop the istallation failing, and you'll have to make
    sure you get the map before VMSINSTAL deletes it whilst tidying up.
    
    Change 2 still might not stop the installation failing later, when it
    finds it hasn't got the formatter image to provide. You could probably
    take this bit out of KIIMAGES:
    
    	$       set file /owner= 'a1$system_uic VMI$KWD:'a1$prefix'FORMATTER.EXE
    	$       write imagelist "A1$FORMATTER ''a1$prefix'FORMATTER.EXE VMI$ROOT:[SYSEXE
    
    to prevent that happening. You'll still need to save the map somehow.
    
    Graham

    GAP,
    
    Couldn't Hans set the A1$NOxxx symbols for at least the Formatter
    image (and maybe the others) to prevent a link during the install
    in the hope that the install might complete and then solve the linking
    problem after?
    
    Regards,
    
    Andrew.D.Wicks 

    
    I don't really understand anything about linking and shareable images,
    but  how about just defining LINK="LINK/MAP" before installing?
    Is that too easy?
    
    But now the customer says they might have an image of their own that
    they are just testing, and it could readily be taken out. He wasn't
    sure about this one, since it belongs to some absent colleague, but 
    if that is the case, I guess we could just do INSTALL>REMOVE and
    then try installing again? 
    
    Hans
    

    I believe VMSINSTAL deletes all your symbols before it does the
    installation so anything involving symbols (or process logicals I
    think) will not work!
    
    Terry

    One of my VMS colleagues have looked into this with the customer,
    they have removed his own images, but still the same problem.
    
    It seems that it complains about LIBRTL which is already at a certain
    adress in VM, LIBRTL is based and is always like that according to
    VMS group. So they say it must be ALL-IN-1 that is doing  something 
    wrong. But this does not seem to be a common problem??
    
    Any more ideas?
    
    Hans 

    Re .6, .8 is right when he says that VMSINSTAL deletes all global
    symbols, which is why I got into my complicated explanation in .5 of
    how to hack round this.
    
    I still think you should do that in order to get the map, since
    although it's not obvious whether this is our problem or VMS's, it
    looks like the link map will be the only way of finding out.
    
    If you hack the link during the installation, you might also want to
    turn off deferred image activiation by setting the appropriate
    (A1$DEFER I think) symbol in A1LINK too. This might well fix the
    problem anyway!
    
    I still can't understand why the main link works, but the formatter
    link fails...
    
    What VMS version is this by the way?
    
    Graham

    I think I pointed this out in another note last night, but here goes. 
    
>    Re .6, .8 is right when he says that VMSINSTAL deletes all global
>    symbols
    
    I believe that local symbols stay behind.  So you could use a local
    symbol if the procedure will look for it.
    
    BTW, WordPerfect messes up FORMATTER.  You have to deassign SMBSRVSHR
    in the SYSTEM group table.  
    
    Ben

    This might be worth trying. The problem with the hacking is that
    they are installing on a system with only a CD-ROM player, no tape
    station. And they don't have enough disk space to copy all the 
    save sets over before installing. And it seems you can't install
    over the net (i.e. put the save sets on another system).
    
    Anyone have any objections to using local symbol LINK = "LINK/MAP"?
    
    Hans

    re.11
>    BTW, WordPerfect messes up FORMATTER.  You have to deassign SMBSRVSHR
    
    I think the installation procedure checks for this one.
    
>    This might be worth trying. The problem with the hacking is that
>    they are installing on a system with only a CD-ROM player, no tape
    
    Try search list logical pointing to "disk:[dir],cd:[a1030.kit]" .
    
    Regards,
    
    Jan

    Sigh...
    
    He has now tried to get link map in two ways, with local symbol
    LINK="LINK/MAP=SYS$LOGIN/FULL" and with VMSINSTAL...OPTIONS M.
    
    Both of them fail when the question about country code comes up,
    suddenly US is not recognized as a valid option.... Any idea why?
    
    (He is now trying the British kit to see if it works better...)
    
    Hans

    re.-1
    
>    LINK="LINK/MAP=SYS$LOGIN/FULL" and with VMSINSTAL...OPTIONS M.
    
    Won't work.
    
    A1LINK.COM links the formatter like this :
    
    $       LINK /NODEBUG /NOTRACEBACK /NOMAP /NOCROSS -
             /EXECUTABLE=a1$work:'a1$prefix'formatter -
                    a1$build:OAFORMATTER, a1$build:OAFORMATTERX,a1$work:OALNM, -
                    a1$work:OAFORMATTER.OPT /option, -
                    a1$build:IMAGEID.OPT /option
    
    Redefining LINK = LINK/MAP is overruled by VMS. Qualifiers
    are evaluated from right to left. The last occurence is taken.
    
    So $LINK /MAP /NOMAP result in no map.
    
    Go for hacking the saveset.
    
    Regards,
    
    Jan
    
    

...Or hack VMSINSTAL's working directory copy of A1LINK.COM after it get restored
to disk from the saveset and BEFORE it gets invoked.

All it takes is a second terminal to SET DEFAULT to SYS$UPDATE:, then SET DEFAULT
to the temporary subdirectory (probably [.A1030]).  Now wait for A1LINK.COM to
be restored and quickly edit it (or replace it with a copy you've got already
edited and waiting somewhere).  It appears that A1LINK.COM (and A1LNKDRV.COM) is
located in A1Lxx030.D, where 'xx' is US on my system and may be a different
country code for you.

And I do believe that .15 meant to say that VMS qualifier are parsed from LEFT
to RIGHT.

/Marty

    Why, oh why...
    
    Well, we have hacked the installation as recommended. But this time
    it stops again at the question for language code. When entering US 
    the response is that it can't find A1LUS30... on the CD. 
    
    As I mentioned sometime earlier he had this problem before, but that
    was when he tried installing with options M or with a symbol LINK
    defined. Even though we could not understand what was going on there,
    we thought the language code problem was somehow related to this.
    
    But now he has rebooted and run the installation. The only difference
    from those earlier times when he got to the linking error, is that now
    he has copied over and hacked the two files, he has not started putting
    them into SYS$UPDATE yet. So I can't see how this could possibly cause
    an error in language code checking. Looks like a separate problem, that
    just shows up arbitrarily???
    
    So, any more bright ideas? Can I put some kind of verify on the install
    procedure to see what is going wrong? 
    
    Thanks,
    Hans

>>    Well, we have hacked the installation as recommended. But this time
>>    it stops again at the question for language code. When entering US 
>>    the response is that it can't find A1LUS30... on the CD. 
    					 ^^^^^^^
    
    	This should be A1LUS030.  Is this just a typo in your note or is
                            ^
    the installation really looking for A1LUS30?
    
    Martin.

    My typo. Should be A1LUS030.
    
    Hans

Well I just did a co-existent installation from CD yesterday with no problems
of any kind.  The co-existant installation is for most practical purposes the
same as a full installation.

/Marty

    Well, the rathole with language code is resolved, it turned up because
    he had mounted the CD before installing. If you do that you must
    mount/system. Or you can just leave it to VMSINSTAL to mount it.
    
    Now on towards the link map again....
    
    Hans

Graham,

ALL-IN-1 v3 is now installed, but still the link problem for 
OA$FORMATTER is there. 

The MAP and a sysgen output can be copied from STKOFF""::

	OA$FORMATTER.MAP;1     1608
	OA$FORMATTER.SYSGEN      37


Best Regrads
LOM

    I'm not really clever enough to understand this. Perhaps you have a
    duff version of LIBRTL? Maybe your process needs more quotas?
    
    Graham.
    
    To avoid everyone else copying the map, here's the bit from the front:
    
      %LINK-E-INSVIRMEM, insufficient virtual memory for 202. pages
            for cluster LIBRTL
      -LINK-E-NOIMGFIL, image file not created
      %LINK-E-INSVIRMEM, insufficient virtual memory for 24. pages
            for cluster XDPS$DPSCLIENTSHR
      -LINK-E-NOIMGFIL, image file not created
    
    
    and the bits about the specified libraries:
    
    Name   Ident   Bytes File                            Creation Date     
    
    LIBRTL V05-001 0     SYS$COMMON:[SYSLIB]LIBRTL.EXE;2 19-JUN-1990 04:38
    XDPS$DPSCLIENTSHR
           DECWRITE V1.0 [SYSLIB]XDPS$DPSCLIENTSHR.EXE;1  3-MAY-1990 15:02
    
    
    Cluster Type Pages Base Addr  Global Sec.       Majorid Minorid
    
    LIBRTL  3    193   00017A00   LIBRTL_001        1       14
            4      8   0002FC00   LIBRTL_002        1       14
            3      1   00030C00   LIBRTL_003        1       14
    XDPS$DPSCLIENTSHR
            3     24   00000200   XDPS$DPSCLIENTSH  2       1000
    
    PS For those on IOSG, the files are in my top directory.