Hi all, Emmanuels proposal seems to be the exact solution. Congrat Emmanuel ;-)
BTW, I also think that ApacheDS should start to use MINA as only a jar dependency. It does not make much sense to build (and test!) MINA again and again while doing a trunks build. MINA is stable and isolated enough to be used only as an external jar. Cheers, On 9/3/06, Alex Karasulu <[EMAIL PROTECTED]> wrote:
Emmanuel Lecharny wrote: > Alex Karasulu a écrit : > >> Hi all, >> >> As you know, we have dependency issues with ApacheDS and MINA which >> need to be resolved. There are several ways in which we can resolve >> the problem. However before I propose some solutions let me restate >> the problem again. Here's some dependencies which exist today: >> >> ApacheDS >> -> MINA filter-codec-asn1 -> shared-asn1 >> -> shared-asn1 >> >> ApacheDS depends on shared-asn1 through two paths as can be seen >> above. One is a direct dependency and the other is via the >> filter-codec-asn1 module in MINA. This makes it so the version of >> shared-asn1 that ApacheDS depends on directly must be the same as the >> version of shared-asn1 that is depended upon by the filter-codec-asn1 >> module. >> >> There are several ways in which we can solve this problem but the key >> is to be able to release this filter-codec-asn1 separately whenever >> changes are made to the shared ASN.1 module. We need to figure out >> how we can release parts of MINA separately to prevent having to push >> a full MINA release every time. > > In my mind, the filter-ASN.1 should not be a part of MINA. MINA has > nothing to do with ASN.1, as it doesn't know how to handle PDUs. > > We should implement a MINA filter into ADS, in order to get rid of this > dependency. Exactly! I fully agree. NOTE: The ASN.1 codec is LDAP specific since it depends on the shared ASN.1 code that is watered down to be a subset of the allowable encodings for BER. This code was originally part of ApacheDS proper. I think Trustin wanted it inside the MINA subproject for critical mass. He moved it there about a year ago. I think we should move it back since ApacheDS is the primary if not only user. Furthermore, I want to prevent this double vote situation whenever we want to release ApacheDS with ASN.1 improvements. Alex
-- Ersin
