libdbi-devel — Discussion about development of the core library and DBD interface

[libdbi-devel] dbi_result_get_int_idx

[<mar...@mh...>]

Hi,

Holger Hetterich writes:
 > "select SUM(length) of read;"
 > 
 > I suppose the result will be integer also. So I do:

I'm not too familiar with PostgreSQL internals, but I wouldn't bet on
that. If the result is a long or long long integer, get_int will return
an error. What does the following call return:

unsigned int attribs;
attribs = dbi_result_get_field_attribs_idx(data, 1);

attribs should contain the size of the integer as explained in dbi/dbi.h
(see the DBI_INTEGER_* definitions).

regards,
Markus

-- 
Markus Hoenicka
http://www.mhoenicka.de
AQ score 38

Re: [libdbi-devel] dbi_result_get_int_idx

[Holger H. <oz...@me...>]

Am Freitag, den 15.04.2011, 20:51 +0200 schrieb Holger Hetterich:
> Hi all,
> 
> I am using the pgsql driver, and run a query for a field "length" in a
> table called "read". Field 'length' has been declared as "integer" in
> postgresql.
> 
> This is what I run through libdbi:
> "select SUM(length) of read;"
> 

sorry that is "select SUM(length) FROM read;

> I suppose the result will be integer also. So I do:
> dbi_result_get_field_type_idx(data,1);
> and the result is 1. AFAIK that's integer.
> 
> Now I do:
> int rr;
> rr = dbi_result_get_int_idx(data,1);
> 
> and it'll be 0. Which is wrong of course.
> 
> Calling the same query on postgresql's console is some 200.000 bytes.
> 
> Calling the dbi error function after this call gives:
> "-7: The requested variable type does not match what libdbi thinks it
> should be"
> 
> If I read strings through libdbi, all is fine, so I am wondering what is
> wrong here.
> 
> Holger
> 
> 
> ------------------------------------------------------------------------------
> Benefiting from Server Virtualization: Beyond Initial Workload 
> Consolidation -- Increasing the use of server virtualization is a top
> priority.Virtualization can reduce costs, simplify management, and improve 
> application availability and disaster protection. Learn more about boosting 
> the value of server virtualization. http://p.sf.net/sfu/vmware-sfdev2dev
> _______________________________________________
> libdbi-devel mailing list
> lib...@li...
> https://lists.sourceforge.net/lists/listinfo/libdbi-devel

[libdbi-devel] dbi_result_get_int_idx

[Holger H. <oz...@me...>]

Hi all,

I am using the pgsql driver, and run a query for a field "length" in a
table called "read". Field 'length' has been declared as "integer" in
postgresql.

This is what I run through libdbi:
"select SUM(length) of read;"

I suppose the result will be integer also. So I do:
dbi_result_get_field_type_idx(data,1);
and the result is 1. AFAIK that's integer.

Now I do:
int rr;
rr = dbi_result_get_int_idx(data,1);

and it'll be 0. Which is wrong of course.

Calling the same query on postgresql's console is some 200.000 bytes.

Calling the dbi error function after this call gives:
"-7: The requested variable type does not match what libdbi thinks it
should be"

If I read strings through libdbi, all is fine, so I am wondering what is
wrong here.

Holger

[libdbi-devel] patches from BalaBit

[<mar...@mh...>]

Peter Czanik writes:
 > Hello,
 > 
 > Syslog-ng uses libdbi for database access. We use an older CVS version
 > and some internal patches. All of these are available in a public git
 > repository:
 > 
 > http://git.balabit.hu/?p=bazsi/libdbi.git;a=shortlog
 > http://git.balabit.hu/?p=bazsi/libdbi-drivers.git;a=shortlog
 > 
 > Some of these patches are specific to syslog-ng, but many of them are
 > also useful for the libdbi community. Part of these patches are already
 > sent upstream, some of them are still only available in our git repo.
 > 
 > Please take a look at our git repos and pick any patches you find
 > useful. If you need further information about a given patch, I'll ask
 > the responsible developer for you.
 > 

Hi,

I understand that you can't use a moving target to build your
Syslog-ng on, but did you ever try a current cvs revision? It'd be
interesting to see what problems still persist.

In any case, I'd like to include as many of these patches as possible,
as this would obviate the need to maintain a separate patched
version. Also, I'm sure your patches will also fix a few problems that
hit other people as well.

I think I've applied some of your requested changes in the past. To
make things easier, would you mind checking recent libdbi and
libdbi-drivers ChangeLogs against your patch list to identify those
who still need my attention? As these ChangeLogs are fairly big, I've
sent them off-list to you.

regards,
Markus
-- 
Markus Hoenicka
http://www.mhoenicka.de
AQ score 38

Re: [libdbi-devel] pgsql test

[<mar...@mh...>]

Peter Czanik writes:
 > One of the main values of libdbi that it hides driver specific details
 > from applications. So I was told by colleagues, that having default port
 > numbers in libdbi would be very welcome.
 > Bye,
 > 

Hi,

thanks for the feedback. I've changed both the mysql and the pgsql
driver to fill in the default ports if none was specified.

regards,
Markus

-- 
Markus Hoenicka
http://www.mhoenicka.de
AQ score 38

[libdbi-devel] patches from BalaBit

[Peter C. <cz...@ba...>]

Hello,

Syslog-ng uses libdbi for database access. We use an older CVS version
and some internal patches. All of these are available in a public git
repository:

http://git.balabit.hu/?p=bazsi/libdbi.git;a=shortlog
http://git.balabit.hu/?p=bazsi/libdbi-drivers.git;a=shortlog

Some of these patches are specific to syslog-ng, but many of them are
also useful for the libdbi community. Part of these patches are already
sent upstream, some of them are still only available in our git repo.

Please take a look at our git repos and pick any patches you find
useful. If you need further information about a given patch, I'll ask
the responsible developer for you.

Bye,

-- 
Peter Czanik (CzP) <cz...@ba...>
BalaBit IT Security / syslog-ng upstream
http://czanik.blogs.balabit.com/

Re: [libdbi-devel] pgsql test

[Peter C. <cz...@ba...>]

Hello,

On 02/19/2011 02:10 PM, mar...@mh... wrote:
> I had to revisit this issue in a different context (libdbi-drivers Bug
> #3163428). Now pgsql and mysql are supposed to deal with ports the
> same way. If a string option is used to set the port, this port is
> used. If no string option is set, but a numeric option is set, use
> that. If neither is set, the port will be set to zero. This
> essentially means that applications using libdbi are expected to
> provide an appropriate port. There is nothing going to happen
> automagically. If this behaviour is not desired, feel free to request
> a change. As mentioned previously, we could provide the well-known
> defaults of pgsql and mysql instead of zero.
>
>   
One of the main values of libdbi that it hides driver specific details
from applications. So I was told by colleagues, that having default port
numbers in libdbi would be very welcome.
Bye,

-- 
Peter Czanik (CzP) <cz...@ba...>
BalaBit IT Security / syslog-ng upstream
http://czanik.blogs.balabit.com/

[libdbi-devel] libdbi and libdbi-drivers patch weekend

[<mar...@mh...>]

Hi all,

I managed to spend a whopping one-and-a-half days trying to deal with
a bunch of patches, bug reports, and feature requests that had piled
up since approx. three months. Unfortunately Sourceforge's
notification message system is still down, so there's no easy way to
keep track of all the changes I've made.

Before you rush to test the latest and greatest, you should be aware
of some changes that may break your builds. There were reports that
the compiler optimizations in the configure scripts do more harm than
good. They also prevented picking up CFLAGS and LDFLAGS settings from
the environment. I've removed all this cruft and made sure that the
regular build target as well as the debug and profile targets just add
the bare minimum to whatever is set in the environment. It is now up
to the package maintainers to add back all optimizations that may have
worked on their platforms. Also, it is possible that builds now fail
on your platform although things went ok in the past. This is usually
caused by the fact that there is no more -I and -L magic in
configure.in. You'll have to fiddle with CFLAGS, LDFLAGS, and CPPFLAGS
until configure stops barfing. E.g. on my platform (FreeBSD 8.1) I
use:

CFLAGS=-I/usr/local/include CPPFLAGS=-I/usr/local/include LDFLAGS=-L/usr/local/lib ./configure --with-mysql --with-pgsql --with-sqlite3 --with-firebird

I'd greatly appreciate if the mailing list was flooded with magic
incantations for all platforms you can think of. I'd be happy to
add them to the INSTALL files or the manuals.

The current status of the drivers that I can test is the following:

MySQL 5.5.7: ok
PostgreSQL 9.0.3: ok
SQLite 3.6.23: ok
Firebird 2.5.0: database is created ok, but test hangs indefinitely
without crashing

I'm inclined to hoof out the firebird driver to avoid further
delays. This would leave only mysql, pgsql, and sqlite3 as 100% tested
drivers at this time.

Please let me know if you think I've missed an important patch,
feature request, or bug report.

regards,
Markus

-- 
Markus Hoenicka
http://www.mhoenicka.de
AQ score 38

[libdbi-devel] [ libdbi-Bugs-3161880 ] [PATCH] Solaris readdir_r semantics patch

[SourceForge.net <no...@so...>]

Bugs item #3161880, was opened at 2011-01-19 19:18
Message generated for change (Settings changed) made by mhoenicka
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=379806&aid=3161880&group_id=23824

Please note that this message will contain a full copy of the comment thread,
including the initial issue submission, for this request,
not just the latest update.
Category: None
Group: None
>Status: Closed
Priority: 5
Private: No
Submitted By: Jan Palus (atler_)
Assigned to: Nobody/Anonymous (nobody)
Summary: [PATCH] Solaris readdir_r semantics patch

Initial Comment:
Default readdir_r on Solaris takes two arguments contrary to POSIX implementation which takes three. This can be changed by defining _POSIX_PTHREAD_SEMANTICS. Instead of explicit define for solaris platform use AC_USE_SYSTEM_EXTENSIONS macro which defines useful variables for running platform (should also define _GNU_SOURCE required by asprintf). Additionally include minimum autoconf version requirement. Patch attached.

----------------------------------------------------------------------

Comment By: Markus Hoenicka (mhoenicka)
Date: 2011-02-20 15:04

Message:
I've applied the patch, see the current cvs revision

----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=379806&aid=3161880&group_id=23824

Re: [libdbi-devel] pgsql test

[<mar...@mh...>]

Peter Czanik writes:
 > And that's right. With MySQL, if I don't configure a port number in
 > syslog-ng.conf, then the default 3306 is used. Setting the port number
 > to another value also works fine.
 > With PostgreSQL if the port is not specified, libdbi tries to connect to
 > port 0 (zero). It works fine, once a port is specified.
 > I was told, that in syslog-ng all databases are treated equal, so it's
 > something in libdbi port handling.

I had to revisit this issue in a different context (libdbi-drivers Bug
#3163428). Now pgsql and mysql are supposed to deal with ports the
same way. If a string option is used to set the port, this port is
used. If no string option is set, but a numeric option is set, use
that. If neither is set, the port will be set to zero. This
essentially means that applications using libdbi are expected to
provide an appropriate port. There is nothing going to happen
automagically. If this behaviour is not desired, feel free to request
a change. As mentioned previously, we could provide the well-known
defaults of pgsql and mysql instead of zero.

regards,
Markus

-- 
Markus Hoenicka
http://www.mhoenicka.de
AQ score 38

Re: [libdbi-devel] next release

[<mar...@mh...>]

Matthew Hall writes:
 > I ran into this openjade issue also.
 > 
 > But I did not try make check.
 > 
 > You probably need the BSD equivalent of docbook-dsssl:
 > 
 > http://packages.debian.org/sid/docbook-dsssl
 > 

I've checked things again on my developer box (FreeBSD 8.1). In order
to transform the docs appropriately, you'll need:

- jade or openjade
- DocBook DSSSL stylesheets to create HTML or TeX output
- TeX and JadeTeX to create PDF output from TeX

Remember that if you build from a tarball, or use a prebuilt package
for your favourite OS, you don't need any of these.

As mentioned previously, we might as well switch the docs to Docbook
XML. However, the requirements would change to:

- xsltproc (the most lightweight XSL engine, but any other would do as
  well)
- DocBook XSL stylesheets to create HTML or fo output
- Java and Fop to create PDF output

Feel free to suggest which requirements you think are easier to
match. But remember that the stylesheets and any Java-based tools are
notoriously hard to check for in configure scripts.

regards,
Markus

-- 
Markus Hoenicka
http://www.mhoenicka.de
AQ score 38

[libdbi-devel] [ libdbi-Bugs-3152613 ] Linux -ffast-math CFLAGS break i386 floating point

[SourceForge.net <no...@so...>]

Bugs item #3152613, was opened at 2011-01-06 21:56
Message generated for change (Comment added) made by mhoenicka
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=379806&aid=3152613&group_id=23824

Please note that this message will contain a full copy of the comment thread,
including the initial issue submission, for this request,
not just the latest update.
Category: None
Group: None
>Status: Closed
Priority: 5
Private: No
Submitted By: clint-fewbar ()
Assigned to: Nobody/Anonymous (nobody)
Summary: Linux -ffast-math CFLAGS break i386 floating point

Initial Comment:
This issue has been reported on numerous Linux distributions:

https://bugzilla.gnome.org/show_bug.cgi?id=611936
https://launchpad.net/bugs/673307
https://bugzilla.redhat.com/show_bug.cgi?id=629964

The fast-math optimization produces broken test results on libdbi-drivers sqlite3 test suite (it "passes" but clearly fails to process the floating point number properly) on i386. The fast-math option should therefore be dropped. Also, arch-specific optimizations should probably be removed in favor of allowing the user to choose their optimizations with CFLAGS.

----------------------------------------------------------------------

>Comment By: Markus Hoenicka (mhoenicka)
Date: 2011-02-19 13:00

Message:
I've removed all platform-specific optimizations and made sure that
user-supplied CFLAGS settings are picked up correctly. Please check the
latest CVS revision to see if it fixes your problem.

----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=379806&aid=3152613&group_id=23824

Re: [libdbi-devel] pgsql test

[Markus H. <mar...@mh...>]

Hi Peter,

Peter Czanik <cz...@ba...> was heard to say:

> With PostgreSQL if the port is not specified, libdbi tries to connect to
> port 0 (zero). It works fine, once a port is specified.
> I was told, that in syslog-ng all databases are treated equal, so it's
> something in libdbi port handling.

Not necessarily. libdbi should not fiddle with port settings at all.  
The drivers do not supply any default setting either if the calling  
application does not explicitly set the port. It might actually be a  
matter of how the client libraries deal with a missing port. I didn't  
check the docs, but MySQL, being a bit more relaxed, may very well use  
a compile-time default in this case, whereas the PostgreSQL lib might  
rather act according to "garbage in, garbage out". In order to make  
the drivers more consistent, we might of course supply a default port  
for those drivers that would otherwise fail.

regards,
Markus


-- 
Markus Hoenicka
http://www.mhoenicka.de
AQ score 38

Re: [libdbi-devel] pgsql test

[Peter C. <cz...@ba...>]

Hello,

On 02/17/2011 09:17 AM, Peter Czanik wrote:
> - test it with syslog-ng on local and network connections, using default
> and non default port numbers 
>   
And that's right. With MySQL, if I don't configure a port number in
syslog-ng.conf, then the default 3306 is used. Setting the port number
to another value also works fine.
With PostgreSQL if the port is not specified, libdbi tries to connect to
port 0 (zero). It works fine, once a port is specified.
I was told, that in syslog-ng all databases are treated equal, so it's
something in libdbi port handling.

-- 
Peter Czanik (CzP) <cz...@ba...>
BalaBit IT Security / syslog-ng upstream
http://czanik.blogs.balabit.com/

[libdbi-devel] pgsql test

[<mar...@mh...>]

Hi Peter,

Peter Czanik writes:
 > Hello,
 > This morning I also added postgresql 9.0 to the mix. "make check" failed
 > at test_pgsql.sh. I was running it as root, and postgresql scripts don't
 > like it. So there was an attempt to su to user postgres. There are two
 > problems with it:
 > - on FreeBSD su is in /usr/bin (not in /bin, as the script expects)
 > - there is no postgres user

The test scripts were donated by the Ubuntu and Debian
maintainers. They still contain some OS-specific stuff. I'm not yet
sure how to deal with this. In contrast to other stuff, the "make
check" stuff should run on any platform without platform-specific
patches. So how do we find out whether or not we have to su to a
particular user in order to do administrative tasks on the server? Is
it only Debian-based OSes that handle postgresql access this way? If
that were the case, we could add some code to ./configure to make sure
this test is run appropriately.

 > So I compiled libdbi-drivers as user czanik, ran make check, and now I have:
 > ==================
 > All 3 tests passed
 > ==================
 > 

Actually, I always build software in my regular account. To the best
of my knowledge, only make install requires root privileges. In any
cases, this should be documented somewhere.

regards,
Markus

-- 
Markus Hoenicka
http://www.mhoenicka.de
AQ score 38

[libdbi-devel] pgsql test

[Peter C. <cz...@ba...>]

Hello,
This morning I also added postgresql 9.0 to the mix. "make check" failed
at test_pgsql.sh. I was running it as root, and postgresql scripts don't
like it. So there was an attempt to su to user postgres. There are two
problems with it:
- on FreeBSD su is in /usr/bin (not in /bin, as the script expects)
- there is no postgres user
So I compiled libdbi-drivers as user czanik, ran make check, and now I have:
==================
All 3 tests passed
==================

Plans for today:
- test all of these on openSUSE 11.4 RC1
- test it with syslog-ng on local and network connections, using default
and non default port numbers (I was told, that using non-default port
numbers is problematic)
Bye,
-- 

Peter Czanik (CzP) <cz...@ba...>
BalaBit IT Security / syslog-ng upstream
http://czanik.blogs.balabit.com/

Re: [libdbi-devel] next release

[Matthew H. <mh...@mh...>]

On Wed, Feb 16, 2011 at 01:27:28PM +0100, Peter Czanik wrote:
> Libdbi: it only compiles if was configured with --disable-docs. Without
> this option, if openjade is missing, it fails because of that, if it's
> there, it still fails, because openjade does not like the document.
> 
> Libdbi-drivers has the same problem with openjade. I compiled it with
> sqlite3 (my primary interest because of syslog-ng) and mysql5.5 (my
> primary db platform). Compilation/installation went fine, but running
> "make check" failed. Should I send the output of "make check" here (it's
> quite long...), or only privately?

Peter,

I ran into this openjade issue also.

But I did not try make check.

You probably need the BSD equivalent of docbook-dsssl:

http://packages.debian.org/sid/docbook-dsssl

Matthew.

Re: [libdbi-devel] next release

[Markus H. <mar...@mh...>]

Peter Czanik <cz...@ba...> was heard to say:

>> SGML is darn old-fashioned, but I hesitate to convert the docs to XML.
>> You'd require Java and FOP to build PDFs then, besides XSL stylesheets
>> and xsltproc. We could as well drop PDFs altogether and build only
>> HTML docs with xsltproc. I'm open to suggestions here. Are xsltproc
>> and the DocBook XSL stylesheets likely to be installed on a
>> run-of-the-mill computer these days?
> No. These were parts of the default Linux desktop install about 7+ years
> ago :-)
>

Are we both talking about XML/XSL here? I readily believe no one  
installs openjade by default these days, but I'd expect to have  
xsltproc around at least. I figured XML is ubiquitous these days.

regards,
Markus


-- 
Markus Hoenicka
http://www.mhoenicka.de
AQ score 38

Re: [libdbi-devel] next release

[Peter C. <cz...@ba...>]

On 02/16/2011 03:42 PM, Markus Hoenicka wrote:
> Hi Peter,
>
> Peter Czanik <cz...@ba...> was heard to say:
>
>> I started testing on FreeBSD 8.1 (the current long term support
>> release). I checked out the latest cvs version for libdbi and libdbi
>> drivers.
>>
>
> That's good as this is my main platform as well. This should make it
> easier for me to reproduce any problems that you run into.
>
>> Libdbi: it only compiles if was configured with --disable-docs. Without
>> this option, if openjade is missing, it fails because of that, if it's
>> there, it still fails, because openjade does not like the document.
>>
>
> Bootstrapping from a cvs checkout is a bit trickier than building from
> a tarball. We decided back then to ship the PDFs and HTMLs along with
> the DocBook sources, so the end-user should never need to build the
> docs. I'll see if I can build the docs on 8.1, but I can't recall ever
> having problems with this as long as the SGML toolchain is complete
> (including DocBook DSSSL stylesheets, an appropriate CATALOG file, and
> a proper $SGML_CATALOG_FILES).
Well, the Makefile only complained about the missing openjade, I did not
install anything else. I guess, I better wait for you here, as I don't
know these tools at all...

> SGML is darn old-fashioned, but I hesitate to convert the docs to XML.
> You'd require Java and FOP to build PDFs then, besides XSL stylesheets
> and xsltproc. We could as well drop PDFs altogether and build only
> HTML docs with xsltproc. I'm open to suggestions here. Are xsltproc
> and the DocBook XSL stylesheets likely to be installed on a
> run-of-the-mill computer these days?
No. These were parts of the default Linux desktop install about 7+ years
ago :-)

>
>> Libdbi-drivers has the same problem with openjade. I compiled it with
>> sqlite3 (my primary interest because of syslog-ng) and mysql5.5 (my
>> primary db platform). Compilation/installation went fine, but running
>> "make check" failed. Should I send the output of "make check" here (it's
>> quite long...), or only privately?
>>
>
> Below is the cause of the failure. Joao, are you listening? This looks
> like a cgreen problem to me. Is it necessary to install cgreen
> separately in order to make this work? Could you please look into this?
>
> --8<---
> FATAL ERROR: Could not find ./bin/my_print_defaults
>
> If you compiled from source, you need to run 'make install' to
> copy the software into the correct location ready for operation.
>
> If you are using a binary release, you must either be at the top
> level of the extracted archive, or pass the --basedir option
> pointing to that location.
> --8<---
Problem solved:

fb81clean# diff test_mysql.sh.old test_mysql.sh
40c40
< mysql_install_db --no-defaults --datadir=${ABSMYTMPDIR} --force
--skip-name-resolve --user=${ME}
---
> mysql_install_db --basedir=/usr/local/ --no-defaults
--datadir=${ABSMYTMPDIR} --force --skip-name-resolve --user=${ME}

Maybe some logic could be added to set "--basedir=" once
"mysql_install_db" is found.

> The sqlite3-specific output of the test looks fine to me.
And with the above fix the mysql tests also run successfully.
==================
All 2 tests passed
==================

>
>> it compiled OK. And I'll also check it on Linux (openSUSE 11.4 RC1)
>>
>
> Much appreciated as I don't have a SuSE system around here.
SuSE is my main desktop system, and I tend to use FreeBSD on servers.
And I was just asked to do some tests on Debian 6.0...

My first syslog-ng tests succeeded, sqlite3 and mysql 5.5 seem to work
OK. I'll do more tests tomorrow and add postresql to the pool, as my
colleagues reported some more problems internally which I want to verify
both with the 0.8.3 release and with the CVS version.

Bye,
-- 

Peter Czanik (CzP) <cz...@ba...>
BalaBit IT Security / syslog-ng upstream
http://czanik.blogs.balabit.com/

Re: [libdbi-devel] next release

[João H. F. <jo...@gm...>]

Hi,

I am listenning.

>> Libdbi-drivers has the same problem with openjade. I compiled it with
>> sqlite3 (my primary interest because of syslog-ng) and mysql5.5 (my
>> primary db platform). Compilation/installation went fine, but running
>> "make check" failed. Should I send the output of "make check" here (it's
>> quite long...), or only privately?
>>

Send the log to me (private).

>
> Below is the cause of the failure. Joao, are you listening? This looks
> like a cgreen problem to me. Is it necessary to install cgreen
> separately in order to make this work? Could you please look into this?
>

I will investigate.



-- 
-----------------------------------------------------------
João Henrique Freitas - joaohf_at_gmail.com
Campinas-SP-Brasil
BSD051283
LPI 1
http://www.joaohfreitas.eti.br

Re: [libdbi-devel] next release

[Markus H. <mar...@mh...>]

Hi Peter,

Peter Czanik <cz...@ba...> was heard to say:

> I started testing on FreeBSD 8.1 (the current long term support
> release). I checked out the latest cvs version for libdbi and libdbi
> drivers.
>

That's good as this is my main platform as well. This should make it  
easier for me to reproduce any problems that you run into.

> Libdbi: it only compiles if was configured with --disable-docs. Without
> this option, if openjade is missing, it fails because of that, if it's
> there, it still fails, because openjade does not like the document.
>

Bootstrapping from a cvs checkout is a bit trickier than building from  
a tarball. We decided back then to ship the PDFs and HTMLs along with  
the DocBook sources, so the end-user should never need to build the  
docs. I'll see if I can build the docs on 8.1, but I can't recall ever  
having problems with this as long as the SGML toolchain is complete  
(including DocBook DSSSL stylesheets, an appropriate CATALOG file, and  
a proper $SGML_CATALOG_FILES). SGML is darn old-fashioned, but I  
hesitate to convert the docs to XML. You'd require Java and FOP to  
build PDFs then, besides XSL stylesheets and xsltproc. We could as  
well drop PDFs altogether and build only HTML docs with xsltproc. I'm  
open to suggestions here. Are xsltproc and the DocBook XSL stylesheets  
likely to be installed on a run-of-the-mill computer these days?

> Libdbi-drivers has the same problem with openjade. I compiled it with
> sqlite3 (my primary interest because of syslog-ng) and mysql5.5 (my
> primary db platform). Compilation/installation went fine, but running
> "make check" failed. Should I send the output of "make check" here (it's
> quite long...), or only privately?
>

Below is the cause of the failure. Joao, are you listening? This looks  
like a cgreen problem to me. Is it necessary to install cgreen  
separately in order to make this work? Could you please look into this?

--8<---
FATAL ERROR: Could not find ./bin/my_print_defaults

If you compiled from source, you need to run 'make install' to
copy the software into the correct location ready for operation.

If you are using a binary release, you must either be at the top
level of the extracted archive, or pass the --basedir option
pointing to that location.
--8<---

The sqlite3-specific output of the test looks fine to me.

> it compiled OK. And I'll also check it on Linux (openSUSE 11.4 RC1)
>

Much appreciated as I don't have a SuSE system around here.

regards,
Markus



-- 
Markus Hoenicka
http://www.mhoenicka.de
AQ score 38

Re: [libdbi-devel] next release

[Peter C. <cz...@ba...>]

Hello,

On 02/10/2011 10:59 PM, mar...@mh... wrote:
> In any case, any kind of help is greatly appreciated to get the new
> release out anytime soon. Test the cvs version on any platform you can
> get hold of. Test the drivers with the latest database engines, as I
> personally can't always live on the bleeding edge. Proofread the docs
> and see whether they reflect the latest changes.
>
>   
I started testing on FreeBSD 8.1 (the current long term support
release). I checked out the latest cvs version for libdbi and libdbi
drivers.

Libdbi: it only compiles if was configured with --disable-docs. Without
this option, if openjade is missing, it fails because of that, if it's
there, it still fails, because openjade does not like the document.

Libdbi-drivers has the same problem with openjade. I compiled it with
sqlite3 (my primary interest because of syslog-ng) and mysql5.5 (my
primary db platform). Compilation/installation went fine, but running
"make check" failed. Should I send the output of "make check" here (it's
quite long...), or only privately?

My next check will be, how it works with syslog-ng. I already see, that
it compiled OK. And I'll also check it on Linux (openSUSE 11.4 RC1)

Bye,

-- 
Peter Czanik (CzP) <cz...@ba...>
BalaBit IT Security / syslog-ng upstream
http://czanik.blogs.balabit.com/

[libdbi-devel] next release

[<mar...@mh...>]

Peter Czanik writes:
 > Hello,
 > 
 > I wonder, when a new release of libdbi and libdbi-drivers could be done.
 > I read in the archives, that there is a new testing framework and many
 > work went in to make it happy. But now all tests are passing. Is there
 > any chance for a release soon?
 > 
 > Background: I help to maintain and upgrade syslog-ng in many different
 > Linux distributions and BSDs. Recently many distributions enabled
 > database (libdbi) support in their syslog-ng package/port. The problem
 > is, that distributions feature the latest release, 0.8.3.  Many of
 > 0.8.3's problems are already fixed in the CVS version.
 > The most notable example is sqlite: support in syslog-ng works only with
 > the CVS version. See
 > http://sourceforge.net/mailarchive/forum.php?thread_name=1217316368.7005.72.camel%40bzorp.balabit&forum_name=libdbi-drivers-devel
 > 
 > Thank you for your help and your great software,

Hi,

I feel your pain, and I wish I could speed things up a bit. I had a
few days off just when Sourceforge was hacked. As we have quite a list
of patches and bugs to work on, I was reluctant to just go ahead
without the ability to check them in one by one. CVS came back online
just today, so I'll try to fix the remaining issues in the small
amount of time that I can currently devote to my software projects.

In any case, any kind of help is greatly appreciated to get the new
release out anytime soon. Test the cvs version on any platform you can
get hold of. Test the drivers with the latest database engines, as I
personally can't always live on the bleeding edge. Proofread the docs
and see whether they reflect the latest changes.

regards,
Markus

-- 
Markus Hoenicka
http://www.mhoenicka.de
AQ score 38

[libdbi-devel] next release

[Peter C. <cz...@ba...>]

Hello,

I wonder, when a new release of libdbi and libdbi-drivers could be done.
I read in the archives, that there is a new testing framework and many
work went in to make it happy. But now all tests are passing. Is there
any chance for a release soon?

Background: I help to maintain and upgrade syslog-ng in many different
Linux distributions and BSDs. Recently many distributions enabled
database (libdbi) support in their syslog-ng package/port. The problem
is, that distributions feature the latest release, 0.8.3.  Many of
0.8.3's problems are already fixed in the CVS version.
The most notable example is sqlite: support in syslog-ng works only with
the CVS version. See
http://sourceforge.net/mailarchive/forum.php?thread_name=1217316368.7005.72.camel%40bzorp.balabit&forum_name=libdbi-drivers-devel

Thank you for your help and your great software,
-- 

Peter Czanik (CzP) <cz...@ba...>
BalaBit IT Security / syslog-ng upstream
http://czanik.blogs.balabit.com/

[libdbi-devel] Fwd from [Gnucash-devel]: Re: I can't find working libdbi version

[Christoph H. <c.h...@gm...>]

Hello there in libdbi-devel !



-------- Original-Nachricht --------
Return-Path: 	<gnu...@gn...>
Delivered-To: 	GMX delivery to c.h...@gm...
Received: 	(qmail invoked by alias); 28 Jan 2011 12:23:46 -0000
Received: 	from code.gnucash.org (EHLO code.gnucash.org) [204.107.200.65] by mx0.gmx.net (mx018) with SMTP; 28 Jan 2011 13:23:46 +0100
Received: 	from code.gnucash.org (code.gnucash.org [127.0.0.1]) by code.gnucash.org (8.14.3/8.14.3) with ESMTP id p0SCMAZe027930; Fri, 28 Jan 2011 07:22:21 -0500
Received: 	from mailout-de.gmx.net (mailout-de.gmx.net [213.165.64.22]) by code.gnucash.org (8.14.3/8.14.3) with SMTP id p0SCM797027922 for <gnu...@gn...>; Fri, 28 Jan 2011 07:22:08 -0500
Received: 	(qmail invoked by alias); 28 Jan 2011 12:22:06 -0000
Received: 	from unknown (EHLO [192.168.2.100]) [89.204.137.70] by mail.gmx.net (mp060) with SMTP; 28 Jan 2011 13:22:06 +0100
X-Provags-ID: 	V01U2FsdGVkX19CrL5pJdxCnC254TCFzOsVSc57daN1VR2GLy9dFG QCb60oki2VcSMY
Message-ID: 	<4D4...@gm...>
Date: 	Fri, 28 Jan 2011 13:26:30 +0100
From: 	Christoph Holtermann <c.h...@gm...>
User-Agent: 	Mozilla/5.0 (X11; U; Linux i686; de; rv:1.9.2.13) Gecko/20101206 SUSE/3.1.7 Thunderbird/3.1.7
MIME-Version: 	1.0
To: 	John Ralls <jr...@ce...>
Subject: 	Re: I can't find working libdbi version
References: 	<4D3...@gm...> <7AF...@ce...> <4D4...@gm...> <768...@ce...> <AAN...@ma...> <4D4...@gm...> <81B...@ce...> <4D4...@gm...> <4D4...@gm...> <4D4...@gm...> <AD4...@ce...>
In-Reply-To: 	<AD4...@ce...>
X-Enigmail-Version: 	1.1.2
X-Spam-Status: 	No, score=0.0 required=5.0 tests=none autolearn=failed version=3.2.5
X-Spam-Checker-Version: 	SpamAssassin 3.2.5 (2008-06-10) on code.gnucash.org
Cc: 	gnu...@gn...
X-BeenThere: 	gnu...@gn...
X-Mailman-Version: 	2.1.11
Precedence: 	list
List-Id: 	GnuCash Software Design and Development Discussions <gnucash-devel.gnucash.org>
List-Unsubscribe: 	<https://lists.gnucash.org/mailman/options/gnucash-devel>, <mailto:gnu...@gn...?subject=unsubscribe>
List-Archive: 	<http://lists.gnucash.org/pipermail/gnucash-devel>
List-Post: 	<mailto:gnu...@gn...>
List-Help: 	<mailto:gnu...@gn...?subject=help>
List-Subscribe: 	<https://lists.gnucash.org/mailman/listinfo/gnucash-devel>, <mailto:gnu...@gn...?subject=subscribe>
Content-Type: 	text/plain; charset="us-ascii"
Content-Transfer-Encoding: 	7bit
Sender: 	gnu...@gn...
Errors-To: 	gnu...@gn...
X-GMX-Antivirus: 	0 (no virus found)
X-GMX-Antispam: 	0 (Mail was not recognized as spam); Detail=5D7Q89H36p5xOp9NadjzZ5MhTPz63H/pEfdV0RMD619L6yJOmtlDcnS/nnWmPd7e+iGQS r6zhL3+CSmJIyok1pE3XoGTdXiZsce+VDYOnkXx7Cn1Un261peDOKUkVLldCnSeTVAZUi9i086pU FrAAHFvtCL+/O0DV1;



Hello !
> Good catch. I've changed the way doubles are serialized in r20177; please give it a try.
>
> Regards,
> John Ralls
>
Thanks ! "make check" runs through without errors.

gnucash still doesn't like libdbi. Log says :
* 13:13:08  WARN <gnc.backend.dbi> [conn_test_dbi_library()] Test_DBI_Library: Double Failed     1,797692e+307 !=      1,000000e+00

Well these two values are surely not the same.

Mysql log is

/usr/sbin/mysqld, Version: 5.1.54-log (SUSE MySQL RPM). started with:
Tcp port: 3306  Unix socket: /var/run/mysql/mysql.sock
Time                 Id Command    Argument
110128 13:15:29     4 Connect   christoph@localhost on gnucash
                    4 Query     CREATE TEMPORARY TABLE numtest ( test_int BIGINT, test_unsigned BIGINT, test_double FLOAT8 )
                    4 Query     INSERT INTO numtest VALUES (-9223372036854775807, 9223372036854775807, 1.7976921348623158e+307)
                    4 Query     SELECT * FROM numtest
110128 13:15:32     4 Query     SHOW TABLES FROM gnucash LIKE 'gnclock'
                    4 Query     BEGIN
                    4 Query     SELECT * FROM gnclock WHERE Hostname = 'Holtermann' AND PID = '2705'
                    4 Query     ROLLBACK
                    4 Quit     
110128 13:15:40     3 Quit     

IT looks like this time only the part BEFORE the decimal separator has been READ. Because the correct value has been stored as far as i understand
the log. ;-)

regards,

Christoph Holtermann
_______________________________________________
gnucash-devel mailing list
gnu...@gn...
https://lists.gnucash.org/mailman/listinfo/gnucash-devel