Re: [Lcd4linux-devel] My first mail, my first plugin, my fisrt SEGFAULT

[Michael R. <re...@eu...>]

Hi Javi,

> I've installed gdb and these are the results:
>=20
> # gdb --args lcd4linux -Fq
> GNU gdb 6.0
> Copyright 2003 Free Software Foundation, Inc.
> (gdb) run
> Starting program: /home/javi/programacion/lcd4linux/lcd4linux -Fqf
>=20
> Program received signal SIGSEGV, Segmentation fault.
> 0x40020995 in mysql_init () from /usr/lib/libmysqlclient.so.12
> (gdb) bt
> #0  0x40020995 in mysql_init () from /usr/lib/libmysqlclient.so.12

> The problem seems to be the mysql_init () function, from /usr/lib/libmy=
sqlclient.so.12

You are right. Strange....

I know nothing about MySQL, but the following from your code looks=20
strange to me:

    MYSQL *conex;
    mysql_init(conex);

conex seems to be something like a 'Context'. As it is a pointer, it=20
points to nowhere, and no memory is allocated for the data structure.

Maybe this should be:

    MYSQL conex;
    mysql_init(&conex);

Note that here the structure is allocated on the stack, and the address=20
is passed to mysql_init().


> =BFIs there something I can do? or is better I forgets the plugin and w=
ait untill the "exec" plugin is ported to the NextGen.

No, I don't want you to forget this plugin. This looks cool! I think=20
there are a lot of uses for such a database plugin!

What would be even cooler are some status informations from MySQL. Any=20
ideas where to get them from?

But I see another problem here, same as with the (not yet ported, for=20
exactly this reason) mail plugin: These plugins can take some time to=20
deliver a result (if, for example, the query is quite complicated, or=20
lots of rows are affected; or if the mailbox must be polled via a slow=20
internet connection). This would distrub lcd4linux's quite tight timings.

Such plugins should work asynchronously. Which in turn is quite complex=20
to implement. The plugin should fork() on initialization, the child=20
would be responsible for fetching the data, the parent takes the results=20
via shared memory (and locked by a mutex) and delivers them to the=20
evaluator.

Or is there another (probably easier) way to achieve this?

Anyone out there with experience with such things?


bye, Michael

--=20
Michael Reinelt                                   Tel: +43 676 3079941
Geisslergasse 4                                   Fax: +43 316  692343
A-8045 Graz, Austria                          e-mail: re...@eu...