Re: [HACKERS] 2-phase commit
От | Christopher Browne |
---|---|
Тема | Re: [HACKERS] 2-phase commit |
Дата | |
Msg-id | m3vfshi2bv.fsf@chvatal.cbbrowne.com обсуждение исходный текст |
Ответ на | Re: [HACKERS] 2-phase commit (Joe Conway <mail@joeconway.com>) |
Список | pgsql-advocacy |
Oops! bruno@wolff.to (Bruno Wolff III) was seen spray-painting on a wall: > On Wed, Aug 27, 2003 at 22:46:58 -0700, > Joe Conway <mail@joeconway.com> wrote: >> >> That's an interesting observation, because I've long thought >> PeopleSoft ought to support Postgres too. From what I recall, their >> database schema is *very* database neutral (at least as of PSFT >> version 7.x) and fairly simple (we ran it on MSSQL 6.5). It would >> probably be pretty easily ported to run on Postgres. > > In my opinion it is too database agnostic. They pretty much just use > the DB as a file. From what I have seen of the system it is one big > hack. Ah, so it's like the way SAP R/3's HR module works. (I expect I'm the only one around that is more than passing familiar with "cluster tables"; quite supremely nonrelational stuff, and quite bletcherous...) To a great extent this comes from the nature of the application. HR is all about collecting together "documents," and these applications replace "paper" with "pseudopaper." > They aren't big on using referential integrity to keep the data > clean. Ditto for SAP R/3; "cleanliness" is, there, imposed by only using their applications to do updates, which includes writing your software to invoke their functions. -- (reverse (concatenate 'string "gro.mca" "@" "enworbbc")) http://www.ntlug.org/~cbbrowne/linuxxian.html ASSEMBLER is a language. Any language that can take a half-dozen keystrokes and compile it down to one byte of code is all right in my books. Though for the REAL programmer, assembler is a waste of time. Why use a compiler when you can code directly into memory through a front panel.
В списке pgsql-advocacy по дате отправления: