3 databases in 3 days hga@xxxxxx (30 Sep 2019 00:36 UTC)
Support for Scheme standards and implementations Lassi Kortela (30 Sep 2019 08:11 UTC)
Re: Support for Scheme standards and implementations hga@xxxxxx (30 Sep 2019 11:25 UTC)
Scheme implementations and portability Lassi Kortela (30 Sep 2019 13:14 UTC)
Re: Scheme implementations and portability John Cowan (30 Sep 2019 19:27 UTC)
Scheme implementations, portability, FFIs Lassi Kortela (30 Sep 2019 21:16 UTC)
Re: Scheme implementations, portability, FFIs John Cowan (30 Sep 2019 22:10 UTC)
JDBC Lassi Kortela (30 Sep 2019 13:15 UTC)
Re: JDBC hga@xxxxxx (30 Sep 2019 13:24 UTC)
Re: JDBC and subprocess protocol Lassi Kortela (30 Sep 2019 14:29 UTC)
Re: JDBC and subprocess protocol hga@xxxxxx (30 Sep 2019 15:16 UTC)
Re: JDBC and subprocess protocol Lassi Kortela (30 Sep 2019 15:47 UTC)
Re: JDBC and subprocess protocol Lassi Kortela (30 Sep 2019 15:55 UTC)
Re: JDBC John Cowan (30 Sep 2019 15:10 UTC)
Re: JDBC Lassi Kortela (30 Sep 2019 15:26 UTC)
Re: JDBC Lassi Kortela (30 Sep 2019 15:34 UTC)
sdbi design in detail and MariaDB CONNECT hga@xxxxxx (30 Sep 2019 16:14 UTC)
Re: sdbi design in detail and MariaDB CONNECT Lassi Kortela (30 Sep 2019 16:28 UTC)
Re: sdbi design in detail and MariaDB CONNECT John Cowan (30 Sep 2019 20:25 UTC)
Re: JDBC John Cowan (30 Sep 2019 16:44 UTC)
Re: JDBC Lassi Kortela (30 Sep 2019 20:52 UTC)
Re: JDBC Alaric Snell-Pym (01 Oct 2019 09:26 UTC)
Re: JDBC hga@xxxxxx (01 Oct 2019 09:55 UTC)
Re: JDBC Alaric Snell-Pym (01 Oct 2019 11:09 UTC)

Re: Support for Scheme standards and implementations hga@xxxxxx 30 Sep 2019 11:25 UTC

> From: Lassi Kortela <xxxxxx@lassi.io>
> Date: Monday, September 30, 2019 3:11 AM
>
> Great work!

Thanks.

>> Thanks to Alex Shinn and his sqlite3 Snow library, and his Chibi
>> Scheme, getting the project up and running separate from a fork of
>> Chibi Scheme took only a day, including creating a couple of cut
>> down from it and customized Gnu Makefiles:
>> https://github.com/Schemepersist/sdbi-chibi
>
> Did you need to patch Chibi, or do the current libraries work with
> stock Chibi and its FFI?

The current libraries all work well with the master branch of Chibi,
and in my experience that means they'll work fine with 0.8.0.  I'm
missing a trick in using the c-link FFI directive, but it's not an
issue with only two FFI connectors so far.

>> Thanks to Takashi Kato for his PostgreSQL wire (net) protocol Snow
>> package....
>
> Takashi maintains a very neat Scheme implementation, Sagittarius,
> that nobody talks about for some reason. It's the only Scheme
> besides Larceny that supports both R6RS and R7RS! He's a very
> responsive maintainer too.  Göran said that he "sometimes fixes bugs
> before you're back from your tea break" and I have a similar
> experience.

It does look very nice, and the last point release was in June:
https://bitbucket.org/ktakashi/sagittarius-scheme/wiki/Home

Chibi Scheme is simpler, follows the truth in advertising principle.
It's small and for me more comprehensible that the other currently
maintained, up to date with R7RS-small Schemes I've looked at like
Sagittarius or Chicken.  Also uses Gnu Make vs. e.g. CMake, which
I'm not familiar with.

> It would be nice to return the favor by implementing the DB API for
> it. R6RS<->R7RS portability is a personal interest mine, so if you
> can ensure the implementation runs solidly on R7RS I can do some
> porting work to R6RS.

I will support R7RS in the fullness of time, and will be happy to
cooperate with R6RS porting.  FFI porting will likely be a bigger
task, depending on how easy it is for an implementation.

Which encourages us to support all the mature databases we can at a
wire (net) protocol level, which includes your subprocess approach.

Which for SQLite3 should work well, and there are advantages to
putting a solid database like that in a separate process, wild
pointer bugs are much less likely to damage your persistent data.

> [ Portable R6RS FFI library. ]
>
> Seems the Postgres library is based on SRFI 106: Basic socket
> interface and has ready wrappers for Chibi and Gauche.

Yes, the cond-expands for the PostgreSQL library are impressive, not
just for Scheme implementations but also for the presence or absence
of some other SRFIs it depends on.  (chibi net) appears to be a
sufficient subset of SRFI 106 that little or nothing was needed there.

>> [ Apache Cassandra ]

> We get more signal :)

When we have a JDBC connector, we'll be able to say the classic line.

- Harold