https://docs.opendap.org/index.php?action=history&feed=atom&title=Length_Bytes 2026-06-18T04:54:31Z Revision history for this page on the wiki MediaWiki 1.38.4 https://docs.opendap.org/index.php?title=Length_Bytes&diff=2236&oldid=prev 2008-06-24T21:28:32Z

<p>New page: The current spec (revision 1.17) contains a note about specifying lengths using a scheme that does not limit the length to one integer. The scheme is proposed for the &lt;nop&gt;BitImage type, b...</p> <p><b>New page</b></p><div>The current spec (revision 1.17) contains a note about specifying lengths using a scheme that does not limit the length to one integer. The scheme is proposed for the &lt;nop&gt;BitImage type, but I wonder if it should be used anywhere we specify a length? The scheme is as follows: the high-order bit is used to indicate whether another length byte follows and the remaining 7 bits are length value. For example, a length of 500 would be 1000 0011 0111 0100 (0x83 0x74). The set high-order bit of the first byte indicates a second byte follows. The cleared high-order bit of the second byte indicates it&#039;s the last byte.<br /> <br /> James Gallagher - 15 Sep 2003<br /> <br /> Take a look at the [http://www.cl.cam.ac.uk/~mgk25/unicode.html UTF-8] FAQ. UTF-8 uses an encoding scheme that is robust. It is easy to spot missing bytes, for example. OTOH, the scheme Tom has suggested has the advantage that it&#039;ll be easy to code and will never use more than N+1 bytes for somethng that would fit in N bytes as an unsigned integer.<br /> <br /> I think we should adopt this unless there&#039;s another widely-used scheme that accomplishes the same goal.<br /> <br /> James Gallagher - 26 Sep 2003</div>

Jimg