Frequently asked questions about the xbase.pm/dbd::xbase modules
This is a list of questions people asked since the module has been announced in fall 1997, and my answers to them.
Jan Pazdziora
You need perl 5.10 or newer. You need \s-1DBI\s0 module version 1.00 or higher, if you want to use the \s-1DBD\s0 driver (which you should).
Yes. It's a standard Perl module so there is no reason it shouldn't. Or, actually, there are a lot of reasons why standard thing do not work on systems that are broken, but I'm trying hard to workaround these bugs. If you find a problem on these platform, send me a description and I'll try to find yet another workaround.
The only possible format in which you can get the date and that the module expect for inserts and updates is a 8 char string '\s-1YYYYMMDD\s0'. It is not possible to change this format. I prefer to do the formating myself since you have more control over it. Because. You get the _DELETED flag as the first value of the array. This gives you a possibility to decide what to do \*(-- undelete, ignore... It's a feature \*(-- you say you want a record of given number, you get it and get additional information, if the record is or isn't marked deleted.
That's correct: DBD::XBase only gives you records that are positively in the file and not deleted. Which shows that XBase.pm is a lower level tool because you can touch records that are marked deleted, while DBD::XBase is higher level \*(-- it gives you \s-1SQL\s0 interface and let's you work with the file more naturaly (what is deleted should stay deleted).
Describe exactly, what you expect and what you get. Send me the file (I understand attachments, uuencode, tar, gzip and zip) so that I can check what it going on and make XBase.pm undestand your file. A small sample (three rows, or so) are generally enough but you can send the whole file if it doesn't have megabytes. Please understand
On Win* platform and with ActiveState port, use ppm to install DBD::XBase from ActiveState's site. You can also just copy the files from the lib directory of the distribution to where perl can find them. Also check whether your make doesn't hide under different names (nmake, gmake). See \*(C`README\*(C'.
Ask your sysadmin to do it for your. If he refuses, fire the sysadmin. See \*(C`README\*(C' for how to install into and use nonstandard place for the module.
For reading \*(-- yes. For writing \*(-- XBase.pm has a locksh and lockex method to lock the file. The question is to what extend Clipper (or Fox* or whatever) uses the same system calls, documentation of native XBase applications doesn't tell this. So the answer is that for multiple updates you should probably consider real \s-1RDBMS\s0 system (PostgreSQL, MySQL, Oracle, to name a few).
No, it doesn't. The character data is returned exactly as it appears in the dbf/dbt file. You probably brought the file from different system that uses differend character encodings. So some bytes in the strings have different meaning on that system. You also probably have fonts in different encoding on that system. In the Czech language, we have about 6 different encoding that affect possition at which accented characters appear. So what you really want to do is to use some external utility to convert the strings to encoding you need \*(-- for example, when I bring the dbf from Win*, it often is in the Windows-1250 or PC-Latin-2 encoding, while the standard is \s-1ISO-8859-2\s0. I use my utility Cz::Cstocs to do the conversion, you maight also try \s-1GNU\s0 program recode or use Text::Iconv Perl module.
Just read the memo field, it will fetch the data from the memo file for you transparently. If you want to match wildcards with DBD::XBase, you have to use \*(C`like\*(C':
select * from table where field like '%str%'
No. At least, I hope no. The software is provided without any warranty, in a hope you might find is useful. Which is by the way the same as with most other software, even if you pay for that. What is different with XBase.pm/DBD::XBase is the fact that if you find out that the results are different from those expected, you are welcome to contact me, describe the problem and send me the files that give troubles to the module, and I'll try to find fix the module.
I try to support any file that looks reasonably as dbf/dbt/fpt/smt/ndx/ntx/mdx/idx/cdx. There are many clones of XBase-like software, each adding its own extension. The module tries to accept all different variations. To do that, I need your cooperation however \*(-- usually good description of the problem, file sample and expected results lead to rather fast patch.
If supports a reasonable subset of the \s-1SQL\s0 syntax, \s-1IMHO\s0. So you can do select, delete, insert and update, create and drop table. If there is something that should be added, let me know and I will consider it. Having said that, I do not expect to ever support joins, for example. This module is more a parser to read files from your legacy applications that a \s-1RDBMS\s0 \*(-- you can find plenty of them around \*(-- use them.
Did you follow the steps in the \*(C`README\*(C' and \*(C`INSTALL\*(C' files? Where did it fail? This module uses a standard way modules in Perl are installed. If you've never installed a module on your system and your system is so non-standard that the general instruction do not help, you should contact your system administrator or the support for your system. Aggregate functions are not supported. It would probably be very slow, since the \s-1DBD\s0 doesn't make use of indexes at the moment. I do not have plans to add this support in some near future. ... but it's there. Is DBD::XBase mad or what? The third part of the first parameter to the connect is the directory where DBD::XBase will look for the dbf files. During connect, the module checks \*(C`if -d $directory\*(C'. So if it says it's not there, it's not there and the only thing DBD::XBase can do about it is to report it to you. It might be that the directory is not mounted, you do not have permissions to it, the script is running under different \s-1UID\s0 than when you try it from command line, or you use relative patch and run the script from a different directory (pwd) than you expect. Anyway, add die "Error reading $dir: $!\n" unless -d $dir; to your script and you will see that it's not DBD::XBase problem.
... why doesn't it read all 10 x n records? Check if the file isn't truncated. \*(C`dbf_dump -i file.dbf\*(C' will tell you the expected number of records and length of one record, like Filename: file.dbf Version: 0x03 (ver. 3) Num of records: 65 Header length: 1313 Record length: 1117 Last change: 1998/12/18 Num fields: 40 So the expected length of the file is at least 1313 + 65 * 1117. If it's shorter, you've got damaged file and XBase.pm/dbf_dump only reads as much rows as it can find in the dbf.
DBD::XBase reads the dbf files directly, using the (included) XBase.pm module. So it will run on any platform with reasonable new perl. With \s-1DBD::ODBC\s0, you need an \s-1ODBC\s0 server, or some program, that \s-1DBD::ODBC\s0 could talk to. Many proprietary software can serve as \s-1ODBC\s0 source for dbf files, it just doesn't seem to run on Un*x systems. And is also much more resource intensive, if you just need to read the file record by record and convert it to \s-1HTML\s0 page or do similary simple operation with it.
XBase.pm doesn't support this directly. You'd probably want to create new table, copy the data and rename back. Patches are always welcome.
Put 'version' => 3 options in to the create call \*(-- that way we say that the dbf file is dBaseIII style.
http://www.adelton.com/perl/DBD-XBase/