Skip to content

Improving DBI

4 messages · Kirill Müller, Paul Gilbert

#
Kirill

TSdbi implements a time series specific API on top of DBI. Some of my 
TSdbi packages use DBI in what you might consider the traditional way 
(TSMySQL, TSPostgreSQL, TSSQLite). TSodbc fudges a bit so it can use 
RODBC, doing some of what would be needed in RODBCDBI. But several of my 
packages interface to non-SQL databases and use only the DBI class 
definitions and a couple of essential generic methods (possibly just 
dbConnect and dbDisconnect). This includes packages TSsdmx, TSmisc, 
TSjson, TSfame, TSbbg, some on CRAN and some not. These packages 
interface to time series data from a variety of sources, many over the 
Internet. They all just wrap other packages in an attempt to standardize 
the API.

I think it would be nice if you can separate the DBI classes and the few 
essential generic methods into a different package from the more SQL 
specific parts of DBI. (I have taken this approach with my packages 
TSdbi and TSsql.)

To get a sense of how I use this in the non-SQL context you might look 
at package TSsdmx, in which the R code has

####### some kludges to make this look like DBI. ######
#for this require("DBI") ; require("RJSDMX")

setClass("sdmxDriver", contains=c("DBIDriver"))

setClass("sdmxConnection", contains=c("DBIConnection", "sdmxDriver"),
    slots=c(dbname="character") )

setMethod("dbConnect", signature(drv="sdmxDriver"),
      definition=function(drv, dbname, ...)
          new("sdmxConnection", dbname=dbname))

# this does nothing but prevent errors if it is called.
setMethod("dbDisconnect", signature(conn="sdmxConnection"),
      definition=function(conn,...) TRUE)

#######     end kludges   ######

Best of luck with your proposal and project.

Happy New Year,
Paul
On 12/30/2015 08:59 PM, Kirill M?ller wrote:
#
Paul


Thanks for your feedback. I'm not sure we want two separate packages for 
DBI, but we can surely split the DBI specification as to make the "SQL" 
part optional. This may mean that more legible error messages are 
returned for virtual methods that are not implemented, or at least that 
the meaning of the current error messages ("unable to find an inherited 
method...") is well communicated. The "kludge" example is in fact a 
proper implementation of the corresponding DBI subset :-)

You could then use the DBI test suite (=specification) just the same, 
simply don't test the "SQL" part of DBI. Would that help?


Best regards

Kirill
On 03.01.2016 23:32, Paul Gilbert wrote:
#
On 01/04/2016 08:50 AM, Kirill M?ller wrote:
For my use it does not make much difference, I can just import what I 
need from DBI. However, it might make a lot of sense if you ever want to 
standardize in layers, for example, if you ever wanted NoSQL to be a 
possible replacement for SQL.

There are different reasons for wanting separate packages, but the 
important one in my mind may not be the one you are thinking about: The 
classes, and the generic methods dbConnect, and dbDisconnect should all 
be extremely stable. On the other hand, the SQL part is likely to go 
through some changes. For sake of discussion let me call the two 
packages DBIclasses and DBIsql. If you make a change in DBIsql my 
packages TSsdmx, TSmisc, and some others, will not be in the upstream 
dependencies, and do not need to be tested for a CRAN submission of 
DBIsql. If DBIclasses and DBIsql are in the one package, DBI, then these 
packages do need to be checked (not just by me but also by you if you 
make an API change and intend to submit to CRAN). These packages in turn 
have a large number of dependencies which can change from time to time 
on their own. Thus things may be broken for reasons having nothing to do 
with your changes, and are beyond your control. Then the CRAN checks 
will fail and your submission will be rejected, or at least require 
considerable additional work. So, it is advisable to avoid having 
dependencies that really can be avoided.
Yes, I hope so. It was only a kludge in the sense that these are not in 
the package I am wrapping, as they are in other packages I wrap like 
RMySQL that use DBI
Yes, I think that might be useful.

Best,
Paul
#
On 04.01.2016 21:42, Paul Gilbert wrote:
Thanks, Paul, I think I got your point. I have opened a GitHub issue for 
further discussion: https://github.com/rstats-db/DBI/issues/72. E-mail 
is fine, too.


-Kirill