[XMLBEANS-481] Base64 to byte[] is not very efficient

• Type: Improvement
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: Version 2.5
• Fix Version/s: Version 5.0.1
• Component/s: DOM
• Assignee: Andreas Beeker
• Reporter: Matej Spiller-Muys
• Created: 03/May/12 20:27
• Updated: 11/Jul/21 22:06
• Resolved: 22/Jun/21 22:38

Description

We are using 30 mb large byte[] string represented as base64 binary type in shema and it is causing out of memory for us on server. Basically converting from base64 string to byte[] is not very efficient.

JavaBase64Holder.java is using: vBytes = v.getBytes("UTF-8") and according to http://www.evanjones.ca/software/java-string-encoding-internals.html it uses base64 size * 4. In our case where there is 30mb large byte[] it uses around 160 mb just for converting string into byte[].

Would it be possible to use v.getBytes("ASCII")? This would reduce the memory usage when converting from string to byte[] by 4 times.
According to http://www.w3.org/TR/xmlschema-2/#base64Binary it should use only ASCII chars anyway.