Forum OpenACS Development: Tried Japanese (and Unicode/multibyte charsets) in OpenACS

I installed openACS 3.2 and made some of my old patches to
convert from Japanese encodings, and it seems to work fine.

I haven't tested the Postgres equivalent of CLOB datatypes yet, but
regular VARCHAR inserts and selects seem to work fine, i.e.,
the unicode support in Postges seems to behave just like the
it does with the Oracle driver, so AOLserver is happy.

So if the OpenACS 4 request processor uses mostly the same code
as the current ACS 4 request processor, then unicode and
multibyte charset handling should work fine, and I have some
patches to help with that.

This is great news, Henry.

In OpenACS 3.2 we simply use the compressed text type rather than the dummied large-object hack I implemented for large binary data.  Users can compile PG with a blocksize up to 32KB and the use of lztext yields text columns that are even larger.  The semantics are exactly the same as for text/varchar.

OpenACS 4x will be based on PG 7.1, which has no limit on the length of a text column.

In other words, it sounds like we're golden once we apply your patches to the ACS 4 request processor.  OpenACS will be using the version included with ACS 4.2 beta - does the beta include your patches?

I don't think the ACS4 release contains the patches, but the
acs-lang package has patches against ACS 4.1 in a little patch file
that is included in the directory, and I don't think ACS 4.2 beta
changed the request processor in any significant way.

Note that there are a couple of small patches required that need to be applied
to AOLserver itself also, which I want to try to get into OpenNSD, since it doesn't look like anyone is going to put them into ArsDigita's
AOlserver release at this point. I'm not sure who to talk to about

Rob Mayoff would be a good person to talk to about getting the patches
into opennsd, but I imagine you've already talked to him about getting
them into the aD version?  Does he dislike the patches for some
reason or is it just that aD's shift towards ACS Java makes AOLserver
less important?
I think it's a combination of the move to Java, and the fact that
i18n was not a priority at arsDigita. I think Rob had no objections, and wanted to merge the
patches in but we never followed up.
What patches exactly are necessary to make multibyte encodings work
with the current 3.2 OpenACS? On the only online resource I found - - there are only patches against ACS

And what version of aolserver is required?

I prepared a set of patches for AOLserver and OpenACS. Look at

It hardcodes support for ShiftJIS.