Re: [HACKERS] Re: MySQL and BerkleyDB (fwd)
Emmanuel Charpentier wrote: Peter Eisentraut wrote: Re: cross-database joins However, I *think* that it could be done by another tool, such as Easysoft's (Nick Gorham's, I think) SQL Engine, which allows for joins between any ODBC-reachable tools. This tool is unreasonably expensive for private use ($800 + $200/year mandatory maintainance). A PostgreSQL alternative would be, IMSAO, a huge benefit, even huger if able to cross-join with ODBC data sources ... M$ Access has this, since version 1, and that's a hell of a handy feature for a lot of cases involving management of multiple data sources You should probably use some front-end tools for most of it (I'd recommend Zope - http://www.zope.org/ ) Or you could try to make something up starting from Gadfly (an all-python SQL databse engine) that is included with zope and also available separately. ... I have to tell that my daily work involves this kind of problems, with data sources ranging from SAS datasets under MVS/XA to Excel files to Oracle databases to younameit ... That's the kind of problem I would *love* to have PostgreSQL to cope with, and *not* M$ Access ... OTOH, much of it could be done if postgres functions could return datasets, which is planned for not too distant future. Hannu
Re: [HACKERS] Re: MySQL and BerkleyDB (fwd)
Peter Eisentraut wrote:
The Hermit Hacker writes:
Is anyone looking at doing this? Is this purely a MySQL-ism, or is it
something that everyone else has except us?
It's not required by SQL, that's for sure. I think in 7.2 we'll tackle
schema support, which will accomplish the same thing. Many people
(including myself) are of the opinion that not allowing cross-db access is
in fact a feature.
I am of the inverse opinion : cross-DB joining is the only reasonable
way to cope with the unfortunate, disgracious, unreasonable, but quite
inescapable real-life fact that all data do not live in the same server
in any but the smallest sites ...
I recently did a plea in this list ("A post-7.1 wishlist") in this
direction, and got an answer (Peter Einstraut ?) that was more or less
on the lines of "over our dead bodies !" ... Sigh ...
However, I *think* that it could be done by another tool, such as
Easysoft's (Nick Gorham's, I think) SQL Engine, which allows for joins
between any ODBC-reachable tools. This tool is unreasonably expensive
for private use ($800 + $200/year mandatory maintainance). A PostgreSQL
alternative would be, IMSAO, a huge benefit, even huger if able to
cross-join with ODBC data sources ...
M$ Access has this, since version 1, and that's a hell of a handy
feature for a lot of cases involving management of multiple data sources
...
why not just build PgSQL, and have transaction support *with* subselects
and everything else that mySQL doesn't have?
I'd *love* to use PgSQL, but it doesn't support cross-DB joins (or at
least I couldn't figure out how to do it.) MySQL handles this, so
I'm using MySQL and would also like to have transaction support...
I have to tell that my daily work involves this kind of problems, with
data sources ranging from SAS datasets under MVS/XA to Excel files to
Oracle databases to younameit ... That's the kind of problem I would
*love* to have PostgreSQL to cope with, and *not* M$ Access ...
[ Back to lurking mode ... ]
E. Charpentier
--
Emmanuel Charpentier
[HACKERS] Re: MySQL and BerkleyDB (fwd)
Is anyone looking at doing this? Is this purely a MySQL-ism, or is it
something that everyone else has except us?
Marc G. Fournier ICQ#7615664 IRC Nick: Scrappy
Systems Administrator @ hub.org
primary: [EMAIL PROTECTED] secondary: scrappy@{freebsd|postgresql}.org
-- Forwarded message --
Date: Mon, 22 Jan 2001 09:03:58 -0600
From: Dave Glowacki [EMAIL PROTECTED]
To: The Hermit Hacker [EMAIL PROTECTED]
Cc: Radovan Gibala [EMAIL PROTECTED], [EMAIL PROTECTED]
Subject: Re: MySQL and BerkleyDB
The Hermit Hacker wrote:
On Mon, 22 Jan 2001, Radovan Gibala wrote:
Is there any possibility to get a port for MySQL with BerkleyDB support?
I realy need the transaction support and I'd like to build MySQL from a
port.
why not just build PgSQL, and have transaction support *with* subselects
and everything else that mySQL doesn't have?
I'd *love* to use PgSQL, but it doesn't support cross-DB joins (or at
least I couldn't figure out how to do it.) MySQL handles this, so
I'm using MySQL and would also like to have transaction support...
Re: [HACKERS] Re: MySQL and BerkleyDB (fwd)
On Mon, Jan 22, 2001 at 11:30:17AM -0400, The Hermit Hacker wrote:
Is anyone looking at doing this? Is this purely a MySQL-ism, or is it
something that everyone else has except us?
Afaik either Informix or Sybase (both?) has it too. Dunno about
anything else.
Marc G. Fournier ICQ#7615664 IRC Nick: Scrappy
Systems Administrator @ hub.org
primary: [EMAIL PROTECTED] secondary: scrappy@{freebsd|postgresql}.org
-- Forwarded message --
Date: Mon, 22 Jan 2001 09:03:58 -0600
From: Dave Glowacki [EMAIL PROTECTED]
To: The Hermit Hacker [EMAIL PROTECTED]
Cc: Radovan Gibala [EMAIL PROTECTED], [EMAIL PROTECTED]
Subject: Re: MySQL and BerkleyDB
The Hermit Hacker wrote:
On Mon, 22 Jan 2001, Radovan Gibala wrote:
Is there any possibility to get a port for MySQL with BerkleyDB support?
I realy need the transaction support and I'd like to build MySQL from a
port.
why not just build PgSQL, and have transaction support *with* subselects
and everything else that mySQL doesn't have?
I'd *love* to use PgSQL, but it doesn't support cross-DB joins (or at
least I couldn't figure out how to do it.) MySQL handles this, so
I'm using MySQL and would also like to have transaction support...
--
marko
AW: [HACKERS] Re: MySQL and BerkleyDB (fwd)
Is anyone looking at doing this? Is this purely a MySQL-ism, or is it something that everyone else has except us? We should not only support access to all db's under one postmaster, but also remote access to other postmaster's databases. All biggie db's allow this in one way or another (synonyms, qualified object names) including 2-phase commit. Ideally this includes access to other db manufacturers, flat files, bdb ... Meaning, that this is a problem needing a generic approach. Andreas Is there any possibility to get a port for MySQL with BerkleyDB support? I realy need the transaction support and I'd like to build MySQL from a port. why not just build PgSQL, and have transaction support *with* subselects and everything else that mySQL doesn't have? I'd *love* to use PgSQL, but it doesn't support cross-DB joins (or at least I couldn't figure out how to do it.) MySQL handles this, so I'm using MySQL and would also like to have transaction support...
Re: [HACKERS] Re: MySQL and BerkleyDB (fwd)
On Mon, Jan 22, 2001 at 11:30:17AM -0400, The Hermit Hacker wrote: Is anyone looking at doing this? Is this purely a MySQL-ism, or is it something that everyone else has except us? Think "schema". I don't foresee supporting joins across multiple databases any time soon (unless someone wants to resurrect the old Mariposa code), but once we have schemas you can put things into different schemas of one database and get most of the user-level benefit of multiple databases, while still being able to join to things that are in other schemas. I haven't looked hard at what it will take to do schemas, but it's high on my priority list for 7.2. Ross Reedstrom has already done some preliminary work ... regards, tom lane
[HACKERS] Re: MySQL and BerkleyDB (fwd)
It's not required by SQL, that's for sure. I think in 7.2 we'll tackle schema support, which will accomplish the same thing. Many people (including myself) are of the opinion that not allowing cross-db access is in fact a feature. I am hoping that when we get to query tree redesign we will have the hooks to do distributed databases etc. Then "cross-db access" will come nearly for free, which the user can choose to use or not. - Thomas
Re: AW: [HACKERS] Re: MySQL and BerkleyDB (fwd)
Zeugswetter Andreas SB writes: It's not required by SQL, that's for sure. I think in 7.2 we'll tackle schema support, which will accomplish the same thing. It does not (e.g. remote access). Maybe this is handled better by an external corba server or some such thing. Many people (including myself) are of the opinion that not allowing cross-db access is in fact a feature. Can you tell me what that "feature" gains you other than mere inconvenience ? And are you implying, that all the other db's are misfeatured in this regard? It's a safety/security measure to me. As long as one backend process can only touch one database you can control things much better. This could be overcome of course, but I'm quite happy with it. -- Peter Eisentraut [EMAIL PROTECTED] http://yi.org/peter-e/
Re: AW: [HACKERS] Re: MySQL and BerkleyDB (fwd)
The Hermit Hacker wrote: sounds like something that should be handled at the application level though ... at least the concept of 'access to other db manufacturers' ... no? If and when we will get functions that can return rowsets (IIRC Oracle's RETURN AND CONTINUE)the simplest case can be easily implemented by having a user-defined method that just does the query for the whole table (or for rows where "field in (x,y,z)" Then putting this in a view and then using it as a table should also be quite simple (or at least possible ;). Only after that should we start optimizing ... -- Hannu
