[Mybusinessbasic-devel] RE: Re: Thoughts on file types and I/O

[Michael S. L. <mi...@ha...>]

Montgomery Tidwell <mailto:MTi...@ca...> wrote:
[snip]
> has the decision been made that PBL will be used as the DB library?
> if not, i suggest you look at the miniDB library
> http://www.atbas.org/minidb/index.php . one thing that i like better
> about miniDB is that it will allow you to create/build/delete indices
> on the fly.    

We were leaning that way but it's not set in stone yet.  Hmmm... Let's
see... The miniDB data files produced are ASCII and can therefore be parsed
using standard unix tools like grep.  That's a nice feature.  I wonder if
there is a speed penalty for it?

My impression is that unless there are performance issues the miniDB library
will work just as well (and perhaps even better than) the PBL library.  Both
libraries use the GPL so licensing is not an issue.

Does anyone else have comments or a preference?

>> 5)  We should add a "DIR=" function to be used in READ statements
>> that will indicate the direction of the read.
> 
> i think you are asking for BBx compatible commands/features, and i
> agree. 

Acutally it's a combination of features from BBx and MAI's OpenBasic.  I'm
not a big BBx user though so if you see something missing from the spec
please mention it.

>> 6)  Fields in Multi-keyed files should be named and access should be
>> permitted by name as well as by field number.  For instance if field
>> number 8 in a file is named "Partnum" then an IOLIST could be
>> written as either "IOLIST #8=PART$" or "IOLIST #PARTNUM=PART$" or
>> more traditionally as "IOLIST *,*,*,*,*,*,*,PART$" 
>> 
>> 7)  Fields in direct files should not be named and access should be
>> permitted only by field number.  For instance if field number 8 in a
>> file contains the "Partnum" information then an IOLIST could be
>> written as either "IOLIST #8=PART$" or more traditionally as "IOLIST
>> *,*,*,*,*,*,*,PART$" 
> 
> are these features that other BB's have, or is this something new?
> the use of "should" comes across rather heavy. i'd rather not see
> this as a requirement. (maybe i'm reading too much into "should" :)  

Again, in an effort to be as broadly compatible as possible, these are a
combination of features from BBx and OpenBasic.

I was saying "should" in an attempt to provoke discussion.  ;-)  Also, to
indicate that nothing was set in stone yet.

>> 8)  The key fields in multi-keyed files can be used in the "KEY="
>> function to specify which of the keys to use for access.  For
>> instance if the "Partnum" field is designated as a key field then a
>> read command could be written as follows "READ
>> (1,KEY=#PARTNUM=PART$)A1$,A2$,A3$..." or as "READ
>> (1,KEY=#8=PART$)A1$,A2$,A3$..." 
> 
> i suggest the use of the BBx syntax of "KNUM=X". where X is generally
> a number, but could be expanded to accept "#PARTNUM". 

Hmmm... if people don't like the OpenBasic format we could use KNUM= and
KNAM= functions instead I suppose.  Comments?

-- 
Michael St. Laurent
Hartwell Corporation