[Info-vax] multiple RMS records in one call (Re: wrong file format)

[Craig A. Berry]

On 1/7/21 9:18 PM, Dave Froble wrote:
> On 1/7/2021 6:48 PM, Craig A. Berry wrote:
>> On 1/7/21 2:35 PM, 1tim.... at gmail.com wrote:
>>> It's a lot easier to create a service that grabs a bunch of records
>>> in a single call than to have to loop and accumulate the data for the
>>> call.
>>
>> So why not use one of the various mechanisms that already do that? E.g.,
>> to get a list of all the usernames on your system, do:
>>
>> $ perl -"MVMS::IndexedFile" -e
>> "tie(%uaf,VMS::IndexedFile,'sys$system:sysuaf.dat',0); print join
>> qq/\n/, keys(%uaf);"
>> APACHE$WWW
>> CRAIG
>> DEFAULT
>> HTTP$NOBODY
>> <snip>
>>
>> All of that could easily be accumulated in an array, and there are a
>> variety of other ways to capture RMS data into whatever data structures
>> you want.  Python no doubt has other mechanisms.  I don't think making
>> it into a system service gets you much over what's already out there.
> 
> I'm not trying to be a smart-ass (yeah, like anyone believes that) but 
> how is perl accessing the data?  I don't know any way to look at the 
> data other than reading through the records.  That is not going to avoid 
> the overhead that Hein mentioned.  Even if mapping the file in a global 
> section, one would still need to break out the records and the username 
> data.  Though that should avoid the overhead of RMS doing the reads.
> 
> Recordsets of lots of data are nice for some tasks.  They sort of aren't 
> much help for other tasks.  When doing updates in each record of a 
> recordset, if exclusive access to the table(s) doesn't exist, updating 
> is non-trivial.

Sure, something is iterating over the records, but that's not something
the developer has to worry about.  Are there more context switches than
one would get with a real RDBMS?  Almost certainly.

But so would there be if the interface requested by the OP were
implemented by DEC/CPQ/HP/HPE/VSI engineering in MACRO or BLISS or C or
whatever, which would also be internally iterating over records. Unless
the interface were limited to fixed-length records, in which case
multiple records could be snagged via IO$_PERFORM or section mapping as
has been mentioned else-thread, but that's not a general solution for
all record formats.