Re: Search OS/400 Physical Files

[CRPence <CRPbottle@xxxxxxxxx>]

  Pete had alluded to a goal of finding a string "trying to reverse 
engineer some 3rd party apps." [I noticed right after I replied, that I 
had snipped that; oops]; possibly a bit unlike the Subject line and text 
from the OP.  An assumption that searching database data would suffice, 
seemed a bit limited.  As such the DASD search was actually offered in 
full sincerity, albeit with the assumption that the idea would likely be 
dismissed without so much as an inquiry.  /Easy/ enough to self-teach by 
creating a file.mbr with a row at the end of over 16MB of data, where 
the row data has a string very likely to be unique.  Dump the dataspace 
and use the search utility; use the details of the dump to [try to learn 
to] navigate, after the string is located.  Given the data were stored 
instead in either an associated space or another object type, alternate 
 techniques and similar learning.  But if such a search is performed 
and found or not found, can help determine if learning or doing more 
might have any value.

Regards, Chuck

vhamberg@xxxxxxxxxxx wrote:
> Chuck you got me rolling on the floor with this one! Trouble is, 
> I love doing this kind of thing!
>
> Now how many civilians know that a context is the MI name for a 
> library!!
>
> And I know I've forgot where the pointers to everything are
> - figured it out once when I was contracted at IBM in 2001.
> Had to see what was actually happening with auxiliary spaces
> and the AXENT stuff. (Just showing off a little - heh!)
>
> CRPence wrote:
>
> > Given a sufficiently [presumed to be effectively] unique string to be 
> > searched, the STRSST D/A/D find function for an address range could be 
> > used to scan the DASD for such data. For occurrences found, visit the 
> > base segment address to determine if it is a dataspace segment, then 
> > find its owning dataspace object, owning member, and then its context. 
> >
> > Pete Hall wrote: 
> > > Well, the other day, I needed to find a table that contained a 
> > > particular ID. Something like: 
> > > grep "ABC123456789" * 
> > > Is it possible to do something like that in qsys.lib? Assume the 
> > > database tables have an unknown schema, and there are a bunch of 
> > > them. <<SNIP>>