Squeak-dev November 2004

squeak-dev@lists.squeakfoundation.org

164 participants
373 discussions

Connecting to Oracle (was: ODBC connection)
by Avi Bryant 29 Nov '04

29 Nov '04

On Mon, 29 Nov 2004 15:40:33 -0500, Welch, Ronald P (US SSA) <ronald.p.welch(a)baesystems.com> wrote: > > Avi, > > I assume pulling the Oracle support over to Squeak from > the VW port would still leave us with the synchronous > connection problem right? That is unless the "proper > Oracle support" was also ported over from as well VW > correct (or written from scratch, of course)? I am assuming > VW does not have the synchronous connection problem. > Is that a valid assumption? Correct on all counts. I don't think porting the Oracle driver from VW is an option, though, for licensing reasons and because I imagine their FFI/plugin API is quite different. So it would have to be from scratch. The OCI protocol isn't published anywhere, is it? It would be great to have a pure-Squeak client like the JDBC "thin" Oracle driver, but it would probably involve some heavy reverse engineering... Avi

3 4

RE: Connecting to Oracle (was: ODBC connection)
by Welch, Ronald P (US SSA) 29 Nov '04

29 Nov '04

John, Thanks for the pointers. BTW, I have looked at GLORP. It is very sophisticated in many ways and is definitely an impressive piece of work. In particular, it goes to great lengths to obviate the impedance mismatch between the relational world and the world of pure objects. It most definitely does excellent job at that. But again, I find the approach taken by ROE of making relations and their algebra into first class objects to be intriguing. I just haven't had much time to play around with ROE as much as I would like because I don't have any Postgress databases laying around that I can easily experiment on. Ron ----------=-=-=-=-=-=-=-=-========oOo========-=-=-=-=-=-=-=-=---------- mailto:Ronald.P.Welch@baesystems.com Phone:(607)770-3701 BAE SYSTEMS Controls 600 Main St Johnson City, NY 13790-1888 ----------=-=-=-=-=-=-=-=-===================-=-=-=-=-=-=-=-=---------- -----Original Message----- From: John M McIntosh [mailto:johnmci@smalltalkconsulting.com] Sent: Monday, November 29, 2004 4:26 PM To: The general-purpose Squeak developers list; avi(a)beta4.com Subject: Re: Connecting to Oracle (was: ODBC connection) If you dig around you'll find the Oracle Call Interface is documented by Oracle in one of their publications. This is what VW uses at the lowest level. It's straight C code binding to atomic values or arrays. This could easily be done in FFI. mmm let's see http://www.csee.umbc.edu/help/oracle8/server.815/a67846/toc.htm The interfacing to Oracle is the easy part. have you looked at Glorp for the OO to relational mapping? On Nov 29, 2004, at 1:08 PM, Avi Bryant wrote: > On Mon, 29 Nov 2004 15:40:33 -0500, Welch, Ronald P (US SSA) > > > <ronald.p.welch(a)baesystems.com> wrote: >> >> Avi, >> >> I assume pulling the Oracle support over to Squeak from >> the VW port would still leave us with the synchronous >> connection problem right? That is unless the "proper >> Oracle support" was also ported over from as well VW >> correct (or written from scratch, of course)? I am assuming >> VW does not have the synchronous connection problem. >> Is that a valid assumption? > > Correct on all counts. I don't think porting the Oracle driver from > VW is an option, though, for licensing reasons and because I imagine > their FFI/plugin API is quite different. So it would have to be from > scratch. > > The OCI protocol isn't published anywhere, is it? It would be great > to have a pure-Squeak client like the JDBC "thin" Oracle driver, but > it would probably involve some heavy reverse engineering... > > Avi > > > -- ======================================================================== === John M. McIntosh <johnmci(a)smalltalkconsulting.com> 1-800-477-2659 Corporate Smalltalk Consulting Ltd. http://www.smalltalkconsulting.com ======================================================================== ===

2 1

RE: Connecting to Oracle (was: ODBC connection)
by Welch, Ronald P (US SSA) 29 Nov '04

29 Nov '04

Hmm, yes I see your point. Oracle is open to a point, but perhaps not open enough in all areas... ----------=-=-=-=-=-=-=-=-========oOo========-=-=-=-=-=-=-=-=---------- mailto:Ronald.P.Welch@baesystems.com Phone:(607)770-3701 BAE SYSTEMS Controls 600 Main St Johnson City, NY 13790-1888 ----------=-=-=-=-=-=-=-=-===================-=-=-=-=-=-=-=-=---------- -----Original Message----- From: Avi Bryant [mailto:avi.bryant@gmail.com] Sent: Monday, November 29, 2004 5:17 PM To: Welch, Ronald P (US SSA) Cc: The general-purpose Squeak developers list Subject: Re: Connecting to Oracle (was: ODBC connection) On Mon, 29 Nov 2004 17:01:52 -0500, Welch, Ronald P (US SSA) <ronald.p.welch(a)baesystems.com> wrote: > Actually, if I were to attempt "going native" with an Oracle > driver I am not sure I would build it on top of OCI. I would > probably attempt to with a pure "thin" driver ala the JDBC > thin drivers using sockets directly in Squeak and eliminating > FFI all together. Yeah, that would be great, but where are the protocol docs? I don't know of any third-party implementations of such "thin" drivers (and only one, for Java, provided by Oracle), so I assume it's not an easy task - compare with Postgres, which must have dozens of client implementations (there are at least 3 for Smalltalk alone). Avi

1 0

RE: Connecting to Oracle (was: ODBC connection)
by Welch, Ronald P (US SSA) 29 Nov '04

29 Nov '04

Actually, if I were to attempt "going native" with an Oracle driver I am not sure I would build it on top of OCI. I would probably attempt to with a pure "thin" driver ala the JDBC thin drivers using sockets directly in Squeak and eliminating FFI all together. (See http://www.lc.leidenuniv.nl/awcourse/oracle/java.920/a96654/overvw.htm) for an overview of the different types of drivers ----------=-=-=-=-=-=-=-=-========oOo========-=-=-=-=-=-=-=-=---------- mailto:Ronald.P.Welch@baesystems.com Phone:(607)770-3701 BAE SYSTEMS Controls 600 Main St Johnson City, NY 13790-1888 ----------=-=-=-=-=-=-=-=-===================-=-=-=-=-=-=-=-=---------- -----Original Message----- From: Avi Bryant [mailto:avi.bryant@gmail.com] Sent: Monday, November 29, 2004 4:52 PM To: The general-purpose Squeak developers list Subject: Connecting to Oracle (was: ODBC connection) On Mon, 29 Nov 2004 13:26:15 -0800, John M McIntosh <johnmci(a)smalltalkconsulting.com> wrote: > If you dig around you'll find the Oracle Call Interface is documented > by Oracle > in one of their publications. This is what VW uses at the lowest level. > It's straight C code > binding to atomic values or arrays. This could easily be done in FFI. > > mmm let's see > http://www.csee.umbc.edu/help/oracle8/server.815/a67846/toc.htm Hmm... but concurrent calls may still be a problem. Good news: their non-blocking interface uses polling rather than callbacks (so would work with FFI). Bad news: " While waiting to retry non-blocking OCI call, the application may not issue any other OCI calls, or an ORA-03124 error will occur. The only exceptions to this rule are OCIBreak() and OCIReset(). See "Cancelling a Non-blocking Call" for more information on these calls. ". So, you won't block the whole VM, but your Oracle queries will be strictly serialized. Avi

2 1

FIX: Making Set/Dictionary etc. loops more robust
by Joshua Scholar 29 Nov '04

29 Nov '04

Changes Set>>do: from do: aBlock tally = 0 ifTrue: [^ self]. 1 to: array size do: [:index | | each | (each _ array at: index) ifNotNil: [aBlock value: each]] to do: aBlock tally = 0 ifTrue: [^ self]. array do: [:each | each ifNotNil: [aBlock value: each]]. The difference is that adding to a Set or Dictionary can put a new rehashed hashed hash table into "array" In the original version, if any block inside of a "do:" or any of the many loops that depend on "do:" called code that added to that Set then there was a chance the set would grow and rehash - and some element would be visited more than once and some not at all. The way it is with this change, adding to a Set in a loop adds an element that might not be visited, and deleting an element doesn't necessarily mean the element won't be visited but the meltdown scenario can't happen. And of course the new version is at least as fast as the old one. One other topic. The fact that Sets (and the many classes derived from Set) never shrink is a potential memory leak. There's probably not any apps that leak this way - after all the fastest way to clear a Set or Dictionary is "aSet_Set new" not "aSet do: [:e| aSet remove: e ifAbsent: [] ]" While this leak may not be important in general memory leaks are probably imporant because many people work by saving an image every day and that's the equivalent to running Squeak forever. Joshua Scholar

1 0

Connecting to Oracle (was: ODBC connection)
by Avi Bryant 29 Nov '04

29 Nov '04

On Mon, 29 Nov 2004 13:26:15 -0800, John M McIntosh <johnmci(a)smalltalkconsulting.com> wrote: > If you dig around you'll find the Oracle Call Interface is documented > by Oracle > in one of their publications. This is what VW uses at the lowest level. > It's straight C code > binding to atomic values or arrays. This could easily be done in FFI. > > mmm let's see > http://www.csee.umbc.edu/help/oracle8/server.815/a67846/toc.htm Hmm... but concurrent calls may still be a problem. Good news: their non-blocking interface uses polling rather than callbacks (so would work with FFI). Bad news: " While waiting to retry non-blocking OCI call, the application may not issue any other OCI calls, or an ORA-03124 error will occur. The only exceptions to this rule are OCIBreak() and OCIReset(). See "Cancelling a Non-blocking Call" for more information on these calls. ". So, you won't block the whole VM, but your Oracle queries will be strictly serialized. Avi

1 0

REMINDER: Harvesting Party Tuesday November 30th 20:00 GMT
by Ken Causey 29 Nov '04

29 Nov '04

Just a reminder that the next Harvesting Party will take place between 20:00 (8 PM) GMT Tuesday November 30th and 00:00 (Midnight) GMT Wednesday December 1st, 2004. As usual we will meet in #squeak on irc.freenode.net. More information about the #squeak IRC channel can be found at http://people.squeakfoundation.org/article/7.html Information about the Harvesting Party idea can be found at http://people.squeakfoundation.org/article/29.html If you have the time please attend and work toward the future of Squeak. The next seven parties are scheduled for 00:00 (12:00 AM) GMT Sunday December 05, 2004 04:00 (04:00 AM) GMT Thursday December 09, 2004 08:00 (08:00 AM) GMT Monday December 13, 2004 12:00 (12:00 PM) GMT Friday December 17, 2004 16:00 (04:00 PM) GMT Tuesday December 21, 2004 20:00 (08:00 PM) GMT Saturday December 25, 2004 00:00 (12:00 AM) GMT Thursday December 30, 2004

1 0

FIX: Set/Dictionary etc. atRandom
by Joshua Scholar 29 Nov '04

29 Nov '04

Ok, I figured out the changeset thing. This does NOT include any change to Set::sizeFor that makes it possible for set capacity to be a power of two. In the case where you have an almost empty Set with a huge capacity, it only calls gcd: a few times anyway (up to 3 commonly). Joshua Scholar

1 0

Connecting to Oracle (was: ODBC connection)
by Welch, Ronald P (US SSA) 29 Nov '04

29 Nov '04

Avi, I assume pulling the Oracle support over to Squeak from the VW port would still leave us with the synchronous connection problem right? That is unless the "proper Oracle support" was also ported over from as well VW correct (or written from scratch, of course)? I am assuming VW does not have the synchronous connection problem. Is that a valid assumption? Ron ----------=-=-=-=-=-=-=-=-========oOo========-=-=-=-=-=-=-=-=---------- mailto:Ronald.P.Welch@baesystems.com Phone:(607)770-3701 BAE SYSTEMS Controls 600 Main St Johnson City, NY 13790-1888 ----------=-=-=-=-=-=-=-=-===================-=-=-=-=-=-=-=-=---------- -----Original Message----- From: Avi Bryant [mailto:avi.bryant@gmail.com] Sent: Monday, November 29, 2004 3:26 PM To: The general-purpose Squeak developers list Subject: Re: ODBC connection On Mon, 29 Nov 2004 14:04:48 -0500, Welch, Ronald P (US SSA) > I was > just wondering if anyone had played around with > Oracle and ROE. Yes, the VW port has Oracle support, which could be pulled over very easily. The problem is finding a reasonable way to connect to an Oracle database from Squeak. Because the ODBC lib uses FFI, it's synchronous - your whole VM will hang while you wait for the results of a query. What I've done in the past is build a simple Java app that accepts queries on a socket, uses the JDBC driver to get results, and then sends them back, but this isn't terribly performant either. A proper Oracle plugin would be great, if anyone has time or money for one... Avi ******************************************************************** This email and any attachments are confidential to the intended recipient and may also be privileged. If you are not the intended recipient please delete it from your system and notify the sender. You should not copy it or use it for any purpose nor disclose or distribute its contents to any other person. ********************************************************************

1 0

FIX atRandom ? how to do changeset?
by Joshua Scholar 29 Nov '04

29 Nov '04

Fix for Set atRandom: I'd make it a changeset but I don't know how to make a changeset with a single entry. I told it to save a ChangeSet and it saved out 26 megabytes. Since Sets never shrink, it does make sense to have the relatively prime scanning for large capacity, nearly empty sets. Note that there is a possible optimization for generating relatively prime numbers. If the Set is a power of two in length, then any odd number will do. Unfortunately Set's class method "sizeFor" makes it impossible to make a set a power of two in length. So, this version checks, and does the optimization if it will do any good. If you want that optimization of an optimazation to be enabled for new Sets then change sizeFor as well. sizeFor: nElements "Large enough size to hold nElements with some slop (see fullCheck)" nElements <= 0 ifTrue: [^ 1]. ^ nElements*2 atRandom: aGenerator | ind entry i span foundrelativeprime | self emptyCheck. entry _ nil. i_1. "try 30 times with random samples" [ entry == nil and: [i<30] ] whileTrue: [ ind _ aGenerator nextInt: array size. entry _ array at: ind. i _ i+1 ]. entry == nil ifTrue: [ "failed 30 random samples - the array must be almost empty" "pick a random, relatively prime span and scan every entry" "Special case for Sets that are powers of two in size. " "If you make a Set that's a power of two in size > 1 then it will" "always be a power of two in size" (array size bitAnd: ((array size) - 1)) = 0 ifTrue: [ span _ (aGenerator nextInt: (((array size) + 1) // 2)) * 2 - 1. ] ifFalse: [ "not a power of two - find any number that's relatively prime" foundrelativeprime _ false. [ foundrelativeprime] whileFalse: [ span _ 1 + (aGenerator nextInt: ((array size)-1)). foundrelativeprime _ (1 = ((array size) gcd: span)). ]. ]. i _ array size. [i > 1] whileTrue: [ ind _ (ind+span-1) \\ (array size) + 1. entry _ array at: ind. entry == nil ifFalse:[^entry]. i _ i-1. ]. self errorEmptyCollection. ]. ^entry.

2 1

← Newer
1
2
3
4
5
6
7
8
...
38
Older →

Jump to page: