Re: SQLITE with Octave

[Przemek Klosowski]

On 05/10/2017 04:02 PM, Andreas Weber wrote:

Am 10.05.2017 um 17:50 schrieb FMourad:

I have browsed all the topics related to the use of sqlite on octave without
finding a clear solution to my situlation. 
According to the following octave wiki :
http://wiki.octave.org/Database_package , two github repositories can be
exploited as sqlite wrappers on octave, but my attemps to use these projects
were unsuccessful. 
These are the links to the repositories : 
•https://github.com/rmartinjak/mex-sqlite3
•https://github.com/Andy1978/octave-sqlite

The second repo linked is mine and is in an very early stage and not
useful at the moment. Perhaps I should add a big fat warning about that.

The main reason that I stopped development was that markuman pointed me
to the mex wrapper (your first linked repo) and I thought reinventing
the weel would be a bad idea.

I see--and I realized that the problem with the MEX-sqlite3 was very superficial.
The provided Makefile does not link all source files---it leaves out structlist.c.
The following builds an extension that works with my Octave  4.0.3 on Linux:
 
mkoctfile --mex -lsqlite3 -o sqlite3.mex sqlite3.c structlist.c

it not only passes its tests but also works with my small database.

One conceptual problem I have is that it returns the data as an array of structs (one per row); e.g. for my simple temperature data,

a(1:5).time,a(1:5).tempc
ans = 2016-04-08 17:30
ans = 2016-04-08 17:45
ans = 2016-04-08 18:00
ans = 2016-04-08 18:15
ans = 2016-04-08 18:30
ans = 35
ans = 35
ans = 36
ans = 36
ans = 36

What's the easiest way to return the entire column of data, i.e. collate a given struct field from every array member into a vector?