tony@oha.UUCP (Tony Olekshy) (02/06/91)
Hello. This is you friendly C-Tree mailing list maintainer. I too am very pleased with the C-Tree ISAM manager. I have used it for a variety of applications, and currently use it to store a `virtual schema' for accessing multiple databases, and some other things they say I can't talk about. C-Tree runs on almost anything, supports variable length records, a very fancy B-Tree+ algorithm, multi-user locking, and an ISAM server that communicates with application clients using IPC. I have used the server with various Unix and with VMS (my DOS code runs single-user to date, although you get various LAN code in the distribution). Note that this is not, apparently, the same server stuff being discussed by other posters to this thread, or else the posting about binary only servers is wrong--more later. This server consideration is important, because the server can buffer data in a mult-iuser environment that would otherwise have to be flushed in order for stand-alone multi-user access to see a consistent database. We see a five times improvement in bulk loads while running multi-user with the server over stand-alone multi-user access. There are tradeoffs here, because writing ISAM code in C is lower level than using some fancy so-called SQL (warning, bias alert), but there is also the very real consideration that I paid once for a reasonable source license, and I can distribute binaries I compile from that source forever. If you're popping out lots of low-priced product, this is critical. Last time I checked (but don't quote me, please), the source license for C-Tree was something like $400 *per* *programmer*, which I find to be an innovative but particularly reasonable arrangement. I do use R-Tree for repetitive reports (such as those on my DD itself), but I have no experience with the other FairCom products. The C-Tree mailing list is, as you might suspect, not associated with FairCom in any way. There are currently ~70 subscribers. You can send mail for the next digest to the address in my signature below, substituting ctree for tony. Requests to be added to the list, and other administration mail, should substitute ctree-request for tony. Opinions expressed in the ctree mailing list are those of the authors of each piece of mail. I'm sure the mailing list would like to know if, and where, I have erred in this posting. -- Yours etc., Tony Olekshy. Internet: tony%oha@CS.UAlberta.CA BITNET: tony%oha.uucp@UALTAMTS.BITNET uucp: alberta!oha!tony Where the spirit does not work with the hand there is no art. --Da Vinci
jhc@irwin.uucp (James H. Coombs) (02/07/91)
In article <472@oha.UUCP> tony@oha.UUCP (Tony Olekshy) writes: >with application clients using IPC. I have used the server with various Unix >and with VMS (my DOS code runs single-user to date, although you get various >LAN code in the distribution). Note that this is not, apparently, the same >server stuff being discussed by other posters to this thread, or else the >posting about binary only servers is wrong--more later. I think the server that you are talking about is the c-tree server. With this server, the client side of the c-tree library sends all low-level requests to the server side. The speed improvement comes from the buffering in the server and, perhaps, from locking records in the process instead of through the OS. The new "Faircom Server" accepts high-level requests from clients. It is multi-threaded. There is a SQL version, which is really an API that translates SQL into native-language requests. I have not seen it, so this brief description is based on my reading only. Faircom verified that licensing is for binaries only, although people should check for themselves. Oh, yes, this server supports transactioning and, I think, some sort of access control. I'm sure there are many more features. --Jim