From: UnixOS2 Archive To: "UnixOS2 Archive" Date: Mon, 30 Jun 2003 14:07:21 EST-10EDT,10,-1,0,7200,3,-1,0,7200,3600 Subject: [UnixOS2_Archive] No. 140 ************************************************** Sunday 29 June 2003 Number 140 ************************************************** Subjects for today 1 Re: Berkeley DB : Andrew MacIntyre **= Email 1 ==========================** Date: Mon, 30 Jun 2003 21:36:30 +1000 (EST) From: Andrew MacIntyre Subject: Re: Berkeley DB On Sat, 28 Jun 2003, Sebastian Wittmeier (ShadoW) wrote: > On Sat, 28 Jun 2003 20:44:51 +1000 (AET), Nicholas Sheppard wrote: > > >Hi all. Does anyone know if there is an OS/2 port of Berkeley DB > >(http://www.sleepycat.com) available from anywhere? I remember I once got > >an older version to compile but never really tested it. I expect I could > >do it again but if someone's aready done it I won't re-do it. > > Ask Henry Sobotka (did some Berkeley DB compiles) and Andrew MacIntyre > (comiles python, which in turn needs Berkeley DB; I believe he wasn't > successful, but could give you some clues). Sorry, been occupied with other things... I have managed to build Sleepycat DB 4.0.14 without too much drama, but I'm not particularly happy with the result as in the context of Python's DB3/4 support, its lacking a bit :-( Specifically, Python's DB3/4 support wants to be able to support the multiuser/multiprocess functionality, and the module's regression tests, or maybe the module itself, don't gracefully handle the lack of this functionality, even though DB itself is easily configured this way. This functionality could be achieved with a SysV shared memory emulation. I know of two such emulations at this point:- - Vincenzo Venuto's daemon process implemention included with his PostgreSQL port; and - a self contained DLL implementation with source but no attribution or licence... Unfortunately the DLL implementation is only an emulation in spirit, and has numerous significant variations in API behaviour from the real thing. I hacked around these, but concluded that the remaining warts were too much. Vincenzo's implementation, while probable quite close to the behaviour of the real SysV SHM, is not OSS, and is (at the moment at least) single thread compatible only - which makes it useless for my purposes with Python :-(. I have asked him to consider releasing this as a separate package with multithread compatibility. While prepared to consider this, he indicated he has higher priorities... It might just be simpler to write my own... -- Andrew I MacIntyre "These thoughts are mine alone..." E-mail: andymac at bullseye.apana.org.au (pref) | Snail: PO Box 370 andymac at pcug.org.au (alt) | Belconnen ACT 2616 Web: http://www.andymac.org/ | Australia