Java API for reading SAS XPT file

Java API for reading SAS XPT file

There isn't. Do a search on the FDA website and you will find (the link to) a specification to the xpt format which is almost worthless, as it is explained for the C-language! Overall, xpt uses a binary format for numbers from the time of the IBM mainframes which is not used by modern computers anymore. In that document, it is not explained how that format works, you will need to do things like byte-shifting and so. Essentially, you are on your own...
I managed to write code for reading and writing xpt, but it costed me always 2 weeks.

Do you understand now why we need to switch to an XML based format?

You can also try to analyze the OpenCDISC source code (I did not use that) to find out how it works.

All data formats have their

All data formats have their benefits and limitations.

Dataset-XML is not mature enough to be used by the industry. Or otherwise, the industry is not ready for Dataset-XML format. Two major problems are

1. File size. According to FDA pilot, Dataset-XML files are 4-5 times larger than the same data in XPT format.

2. Dataset-XML data requires a presence of define.xml file. Otherwise it's not usable.

At this point I like SAS XPT format much better and do not see clearly why do we need to switch to XML based format? 

Kind Regards, 

Sergiy

SAS XPT

Come on Sergiy!

As a Java developer you should know that file size is not important! Memory usage is! If you use a good parsing method, the memory usage of parsing XML or XPT is exactly the same.
And have you ever zipped a Dataset-XML file? You can zip it to a file size that is less than 10% of the size of your XPT file. Modern software can read these zipped XML files without needing to unzip first.

The industry is SCREAMING for an XML based format.The Japanese, who are far ahead of us, want to get rid of SAS-XPT as soon as possible. They really LOVE Dataset-XML.

And yes, you need a define.xml file - that's exactly about modern IT: separation of data and metadata. And nobody forbids you to pack your define together with your Dataset-XML file.

When talking to my colleagues at sponsor companies, they would like to switch to XML as soon as possible. Many have a process in place that generates SDTM in some XML and then "downgrade" it to SAS-XPT.
The industry is really ready for Dataset-XML, but maybe the FDA isn't.

With best regards,

Jozef

"As a Java developer you

"As a Java developer you should know..."

I am not java developer. I am a statistical programmer. And as a user of clinical data I prefer XPT format over Dataset-XML. For me it's not a matter of academic exercise, but rather real everyday work.

I never heard that "industry is SCREANING for XML format". Otherwise, all my colleagues complain about it.

Sorry but we have very different experience on this subject.

Kind Regards, 

Sergiy