Re: How to retrieve the data compression method of a save file?

On 11-Feb-2016 08:18 -0700, rob wrote:

On 11-Feb-2016 08:01 -0700, Szepesi Frigyes wrote:

On 10-Feb-2016 06:11 -0700, Vernon Hamberg wrote:

To the original poster - there has to be something in the SAVF
that would tell you this. SAVFs can be read record by record, and
there might be something in the OIR of the object. <<SNIP>>

I tried to dump 3 save files I created with different DTACPR
settings. Dumps do not tell me anything,
but I also copied those save files to the IFS with CPYTOSTMF
command. In the files with *MEDIUM and *HIGH I saw two suspicious
strings at the same position:L D_TRS (for *MEDIUM,uses TERSE
algorythm) and LD_LZ1 (for *HIGH, uses LZ1 algorythm) I could not
see anything for files without or with LOW compression.

I wonder if IBM would create a new SQL based "IBM i Services",
"Storage Services" for this? Perhaps QSYS2.SAVE_FILE_ATTRIBUTES?
That column would have values for:
- No compression
- Low compression
- Medium compression
- High compression
and they could use internal APIs to retrieve the inforation you're
seeing in DMPOBJ.
<<SNIP>>

The Dump Object (DMPOBJ) does not dump any of the Dump Space data records. The OP used the Copy To Stream File (CPYTOSTMF) to get the binary data /records/ from the Dump Space into a Stream File (STMF) [though the data could have been copied to a database file or to a User Space (USRSPC) instead].

IMO the more appropriate approach would be to ask that the List Save File (QSRLSAVF) API be updated to include the type of compression; anyone could then create their own tooling, and that which is not limited only to what the SQL [services] might provide.

Because the copied data can be effectively disassembled\reverse-engineered [others have already done much of that already over the years, so examples may exist on the web], knowing exactly what offset to look for the compression-type could probably be found predictably; of course fully dependent upon the implementation by the OS which is subject to change [and why getting the info from the appropriate API is more, IMO most, appropriate].