[
Lists Home |
Date Index |
Thread Index
]
Robin Berjon wrote:
> what intrigues me most is not the absence of open source
implementations,
> but rather the absence of open source projects around ASN.1 that
implement
> at least a sizeable chunk of the ASN.1 body of specifications and/or
> have a sufficiently active community working towards that goal.
As I mentioned in an earlier mail, much of the internet
development that goes on these days is focused on the specific needs
of some interest group or protocol, rather than the development of
core technologies. This is seen clearly in the many open source
implementations of ASN.1 subsets. It turns out that almost each of
these projects was implementing ASN.1 or an associated encoder/decoder
in order to accomplish some specific application related goal. Since
accomplishing the goals of any one project always requires only a
subset of full ASN.1, what we have is many partial implementations.
Thus, we've got OCLC libraries for encoding/decoding BER in the Z39.50
community -- but no OCLC ASN.1 compiler or any of the other encoders
since they aren't needed by Z39.50. The SNMP community did the same
thing, their systems need a limited subset of ASN.1 encoding and
that's all they could justify writing. You'll also find, for instance,
quite number of open source packages that can only handle the limited
subset required by X.500 certificates. All but a few of the "Open
Source ASN.1" projects have been doing ASN.1 as a means to an end
rather than an end in itself. Unfortunately, the projects that had
grander goals tended to run out of steam back in the early 90's --
before open source became as important as it is today -- as OSI lost
favor and as it became obvious that implementors wouldn't be able to
compete with the high quality commercial implementations.
The problems ASN.1 has had are, I think, largely due to when
it was defined and the context in which it was defined. ASN.1 came out
in a world where Open Source was not important and by the time Open
Source became important, ASN.1 wasn't "cool" any more. Few people
could understand the value of fully implementing a system when they
could only identify needs for a subset. Also, ASN.1 was defined in the
OSI world... Of course, the world is very different today. Today, we
have the XML movement which has revived all the religion of generic
data encoding and interchange that was so important to us back in the
80's when ASN.1 first was defined. And, the XML movement has shown
people the value and "coolness" of working on general framework
components... (XML is, on its own, beginning to approach the
complexity and breadth of OSI...) Today, it makes more sense to
propose an Open Source ASN.1 than it has for years. Not for technical
reasons, but for political or social reasons.
Nonetheless, my goal in starting this thread wasn't to argue
for an Open Source project but rather simply to bring awareness to the
fact that ASN.1 should be considered just as much an "XML Schema
Language" as any other. Also, to point out that ASN.1 provides a
solution to at least a large number of the requirements for "binary
XML." Given that awareness, there are many paths that can be taken.
bob wyman
|
