Thread: hacked unixodbc driver..
Hello, I hacked the unix odbc driver that comes with the postgresql 6.5 snapshots and incorporated the GetPrivateProfileString Parsing functions from the IODBC driver manager (libiodbc-2.50.2). The current unix odbc driver that comes with postgresql has a very buggy GetPrivateProfileString and it dumps core from time to time, plus it's not too compatible with tabs and nulls on fields which the libiodbc GetPrivateProfileString doesn't have a problem with. It also incorporates the latest odbc src (windows).. I've used this driver with Solaris X86 2.7 / RedHat linux 6.0 without any problems... It also works great with PHP 3.0.9 and DBD-ODBC-20.. :) This way I can run the same php scripts on both windows and linux/solaris platforms.. :) you will need http://www.openlinksw.com/iodbc and download the latest iodbc driver manager.. (2.50.2) compile and install it, make sure it puts it's odbc.ini in /etc grab the hacked unix odbc driver from http://www.odc.net/~gdaswani/sources/odbc.tar.gz tar -zxvf the file on your ~postgresql-snapshot src/interfaces dir.. The remake postgres with ./configure --prefix=FILLINPREFIX --with-odbc --with-odbcinst=/etc an example odbc.ini file. /etc/odbc.ini ; ; ; [ODBC Data Sources] Openlink = Openlink (MT) template1 = template1 [template1] Driver = /opt/pgsql/lib/libpsqlodbc.so Description = PostgreSQL Template1 Database Servername = localhost Username = Password = Database = template1 Protocol = 6.4 Options = ReadOnly = no Debug = 0 Comm = 1 (If Debug = 1, puts debugging info on /tmp and if Comm = 1, puts debugging info on /tmp).. George Daswani
> I hacked the unix odbc driver that comes with the postgresql 6.5
> snapshots and incorporated the GetPrivateProfileString Parsing functions
> from the IODBC driver manager (libiodbc-2.50.2).
The iODBC code was submitted to Postgres way too close to the release
date to be incorporated into our code tree. Can we restart the
discussion on this in a few days, after the v6.5 release cycle has
settled down? Once we know what we are going to do, then we can post
patches on the plain v6.5 release and update the cvs tree to reflect
both Peter Harvey's and your contributions, and to make sure we are
back in sync with Byron's non-cvs code tree.
Regards.
- Thomas
--
Thomas Lockhart lockhart@alumni.caltech.edu
South Pasadena, California
is that nobody who hack the ODBC driver know how to make diffs files ? This is getting annoying, first the guy (from UnixODBC, I think) who introduces itself on this list saying: "Hey, I want write cvs access on the tree or I will make a fork of the odbc driver ..." without even asking who is the driver mantainer, etc ... very, very polite ... Now, there is Daswani, who say (more or less): "I've hacked the odbc driver, you want the changes ?, download this..." (the whole driver) Please, people, this is not how the Open Source work, there must be a _little_ more cooperation, if anybody will make their own with the driver, things will go crazy, really crazy. Just a: "Hi, I've hacked the driver, the changes are ..., here is the patch." would be excellent. btw, I would like to see Byron (the current maintainer) to say something about their situation with respect to the driver (last time he was a little complicated with their personal situation, hope he is better now). Sergio
Hi, Relax. OpenSource maintainers can simply say that we only accept patchs and we only give out cvs write access to those who have submitted some quality patchs. Thomas said as much to me. Any developer who feels that creating a code fork is easier than dealing with the maintainer than they should go ahead and do so. I say this, not because I like code forks, but because I know of projects (this is not one of them) which have maintainers who continue to control code long past when they are really able to commit any time to do the job. re. unixODBC I prefer to 'mirror' some driver source. This way I can make any changes and enhancements to ensure it meets some tests (ie works with StarOffice) and ensure that it is availible to unixODBC users while I wait for changes to be accepted by the source maintainers. The problem with ODBC is that nonone has been handling it holistically... at least no OpenSource developers and this is exactly what is required to get ODBC moving on UNIX/Linux. This is why I started unixODBC. The plan is to be less holistic when unixODBC 1.0 comes out. In other words; the focus will return to things like the DM and drivers will go through compliance tests. The compliance tests will be sent to driver source maintainers and they can do what they want with the info... unixODBC developers will not try to fix all drivers out there. Having said all of that... I have offered to do the patch submissions when asked (ie when PostgreSQL folks are ready). Also note; a number of iODBC and unixODBC developers are currently working on actual code changes and enhancements to get ODBC into KDE. This should be very healthy and will likely see something of a merge between the two. Peter Harvey BTW: Please do not discourage George, or any other developer, from spending time on PostgreSQL or any other project. Developers such as George deserve some lee-way because they spend a great deal of their time on this stuff for free. I often have to remind myself that I am working with a group of volunteers... not paid employees... it always changes the tone of my emails. > -----Original Message----- > From: owner-pgsql-interfaces@postgreSQL.org > [mailto:owner-pgsql-interfaces@postgreSQL.org]On Behalf Of Sergio > Kessler > Sent: Saturday, January 08, 2000 1:53 PM > To: George Henry C. Daswani > Cc: pgsql-interfaces@postgreSQL.org > Subject: Re: [INTERFACES] hacked unixodbc driver.. > > > > is that nobody who hack the ODBC driver know how to make > diffs files ? > > This is getting annoying, first the guy (from UnixODBC, I think) > who introduces itself on this list saying: > > "Hey, I want write cvs access on the tree or I will make > a fork of the odbc driver ..." > > without even asking who is the driver mantainer, etc ... > very, very polite ... > > Now, there is Daswani, who say (more or less): > > "I've hacked the odbc driver, you want the changes ?, > download this..." (the whole driver) > > Please, people, this is not how the Open Source work, there > must be a _little_ more cooperation, if anybody will make their > own with the driver, things will go crazy, really crazy. > > Just a: > > "Hi, I've hacked the driver, the changes are ..., here is > the patch." > > would be excellent. > > btw, I would like to see Byron (the current maintainer) to say > something about their situation with respect to the driver > (last time he was a little complicated with their personal > situation, hope he is better now). > > Sergio > >
It would be great to make diff files and give it to the postgresql maintainers but at this point, I really wouldn't want this driver to be used and entered into the cvs without testing.. It works great for me, and let's me run the exact same php and perl dbd-odbc code on windows / linux / and solaris.. Steve Harvey is working on his excellent unixODBC package and I believe it's great, in the meantime it doesn't work too well with solaris and when the time comes that it works - it would be nice that it get's incorporated into the postgresql source tree.. At this time, the driver hack works okay for me, and doesn't stop me from using postgresql with odbc - when the time comes that unixodbc works on solaris, i'll be a happy camper. In the meantime, I surely won't lodbc, and dbd-odbc. I'm pretty sure there's other people out there who would like the same.. I'd rather not use dbd-pg or php-pg because I also use the exact same code on an informix database using openlink odbc..et a broken GetPrivateProfileString parser from the current postgresql snapshot stop me from using php- George Daswani.. > > is that nobody who hack the ODBC driver know how to make > diffs files ? > > This is getting annoying, first the guy (from UnixODBC, I think) > who introduces itself on this list saying: > > "Hey, I want write cvs access on the tree or I will make > a fork of the odbc driver ..." > > without even asking who is the driver mantainer, etc ... > very, very polite ... > > Now, there is Daswani, who say (more or less): > > "I've hacked the odbc driver, you want the changes ?, > download this..." (the whole driver) > > Please, people, this is not how the Open Source work, there > must be a _little_ more cooperation, if anybody will make their > own with the driver, things will go crazy, really crazy. > > Just a: > > "Hi, I've hacked the driver, the changes are ..., here is > the patch." > > would be excellent. > > btw, I would like to see Byron (the current maintainer) to say > something about their situation with respect to the driver > (last time he was a little complicated with their personal > situation, hope he is better now). > > Sergio >
It would be great to make diff files and give it to the postgresql maintainers but at this point, I really wouldn't want this driver to be used and entered into the cvs without testing.. It works great for me, and let's me run the exact same php and perl dbd-odbc code on windows / linux / and solaris.. Steve Harvey is working on his excellent unixODBC package and I believe it's great, in the meantime it doesn't work too well with solaris and when the time comes that it works - it would be nice that it get's incorporated into the postgresql source tree.. At this time, the driver hack works okay for me, and doesn't stop me from using postgresql with odbc - when the time comes that unixodbc works on solaris, i'll be a happy camper. In the meantime, I surely won't lodbc, and dbd-odbc. I'm pretty sure there's other people out there who would like the same.. I'd rather not use dbd-pg or php-pg because I also use the exact same code on an informix database using openlink odbc. George Daswani.. > > is that nobody who hack the ODBC driver know how to make > diffs files ? > > This is getting annoying, first the guy (from UnixODBC, I think) > who introduces itself on this list saying: > > "Hey, I want write cvs access on the tree or I will make > a fork of the odbc driver ..." > > without even asking who is the driver mantainer, etc ... > very, very polite ... > > Now, there is Daswani, who say (more or less): > > "I've hacked the odbc driver, you want the changes ?, > download this..." (the whole driver) > > Please, people, this is not how the Open Source work, there > must be a _little_ more cooperation, if anybody will make their > own with the driver, things will go crazy, really crazy. > > Just a: > > "Hi, I've hacked the driver, the changes are ..., here is > the patch." > > would be excellent. > > btw, I would like to see Byron (the current maintainer) to say > something about their situation with respect to the driver > (last time he was a little complicated with their personal > situation, hope he is better now). > > Sergio >
> windows / linux / and solaris.. Steve Harvey is working on his excellent > unixODBC package > and I believe it's great, in the meantime it doesn't work too well with > solaris and when the time comes that it works - it would be nice that it > get's incorporated into the postgresql source tree.. George, Hang tight on Solaris... the plan was to wait until unixODBC 1.0 (on Linux) before making the minor changes required to get it to work on Solaris etc. I have a Solaris/Sparc and a Solaris/Intel install here for this reason. Its probably going to happen sooner now that unixODBC is making its way into KDE 2.0. Thanks for your insight! Peter Harvey BTW: Nick has uploaded some fixs for you at... ( ftp://ftp.easysoft.com/pub/beta/unixODBC/unixODBC-DM-1.0.8.tgz ). Please let me know if this does not fix your php problems on Linux. Also note; we are having problems with the unixODBC cvs AND we are trying to get this stuff into KDE... so we may be a bit delayed in getting quick fixs out over the next week or two.