- Pharo-dev - lists.pharo.org

Re: [Pharo-dev] [Pharo-users] [Ann] The Garage database driver API

by Santiago Bragagnolo

Cool! :D On Thu, 7 May 2015 at 11:37 Guillermo Polito <guillermopolito(a)gmail.com> wrote: > Hi all, > > I'm pleased to announce the release 0.1 of Garage, a database driver API. > Garage is meant to be a common interface to access several database > servers, in a JDBC style. > > The current version works in Pharo3 and Pharo4 and includes drivers for: > - mysql (pure smalltalk) > - postgres (pure smalltalk) > - sqlite (using ffi) > - opendbx (allowing connecting with several db vendors, using ffi) > > We have a nice test suite (that maybe you can help to enlarge) that is > tested every day in Jenkins [1]. > > Also, in garage's website [2] you'll find > - installation instructions > - basic usage > - documentation of features such as statements, transactions, cursored > fetches. > - issue tracker, backlog and other information > > Feedback (and contributions) are very welcome! > Enjoy, > Guille > > [1] https://ci.inria.fr/dbxtalk/view/Garage/ > [2] http://guillep.github.io/DBXTalk/garage/ >

May 7, 2015

[Ann] The Garage database driver API

by Guillermo Polito

Hi all, I'm pleased to announce the release 0.1 of Garage, a database driver API. Garage is meant to be a common interface to access several database servers, in a JDBC style. The current version works in Pharo3 and Pharo4 and includes drivers for: - mysql (pure smalltalk) - postgres (pure smalltalk) - sqlite (using ffi) - opendbx (allowing connecting with several db vendors, using ffi) We have a nice test suite (that maybe you can help to enlarge) that is tested every day in Jenkins [1]. Also, in garage's website [2] you'll find - installation instructions - basic usage - documentation of features such as statements, transactions, cursored fetches. - issue tracker, backlog and other information Feedback (and contributions) are very welcome! Enjoy, Guille [1] https://ci.inria.fr/dbxtalk/view/Garage/ [2] http://guillep.github.io/DBXTalk/garage/

May 7, 2015

Re: [Pharo-dev] NativeBoost and memory management

by Luc Fabresse

2015-05-07 9:58 GMT+02:00 Sven Van Caekenberghe <sven(a)stfx.eu>: > > > On 07 May 2015, at 08:53, Holger Freyther <holger(a)freyther.de> wrote: > > > > I used NB to access openlog, closelog and syslog of libc > > Why not use UDP datagrams to send messages to syslog ? > such as what Olivier did: http://smalltalkhub.com/#!/~olivierauverlot/Syslog Luc > > If you do it manually it should be just a couple of lines of code. > > Or have a look at SysLogSender in SystemLogger on StHub. > > No need to mess with native code, unless you want to, of course. > >

May 7, 2015

Re: [Pharo-dev] NativeBoost and memory management

by Sven Van Caekenberghe

> On 07 May 2015, at 08:53, Holger Freyther <holger(a)freyther.de> wrote: > > I used NB to access openlog, closelog and syslog of libc Why not use UDP datagrams to send messages to syslog ? If you do it manually it should be just a couple of lines of code. Or have a look at SysLogSender in SystemLogger on StHub. No need to mess with native code, unless you want to, of course.

May 7, 2015

Re: [Pharo-dev] RewriteTool

by phil＠highoctane.be

Maybe this could be linked to some mongotalk issues with connections in the pools. I've had some in my app and had to make sure all handles were right before a lot of the calls. Phil On Thu, May 7, 2015 at 9:35 AM, Jean-Christophe Bach < jeanchristophe.bach(a)inria.fr> wrote: > * phil(a)highoctane.be <phil(a)highoctane.be> [07.05.2015. @09:28:45 +0200]: > > > Nope, not found. Even using the search box on STH > > > > Phil > > Well, I have just tried andâ¦ I think it is a "SchrÃ¶dinger project": it > is available and not available. > Few minutes ago, it was available, now not anymore. > > JC >

May 7, 2015

Re: [Pharo-dev] RewriteTool

by Jean-Christophe Bach

* phil(a)highoctane.be <phil(a)highoctane.be> [07.05.2015. @09:28:45 +0200]: > Nope, not found. Even using the search box on STH > > Phil Well, I have just tried andâ¦ I think it is a "SchrÃ¶dinger project": it is available and not available. Few minutes ago, it was available, now not anymore. JC

May 7, 2015

Re: [Pharo-dev] RewriteTool

by phil＠highoctane.be

Nope, not found. Even using the search box on STH Phil On Thu, May 7, 2015 at 9:23 AM, Jean-Christophe Bach < jeanchristophe.bach(a)inria.fr> wrote: > * Sean P. DeNigris <sean(a)clipperadams.com> [06.05.2015. @12:26:49 -0700]: > > > Sean P. DeNigris wrote > > > http://smalltalkhub.com/#!/~MarkRizun/RewriteTool/ , which is now > about 70 > > > commits ahead of the first! > > > > Alarmingly, Mark's repo seems to only be intermittently available. Many > > times sthub returns 404... > > > > ----- > > Cheers, > > Sean > > For that, it is the trailing / > => http://smalltalkhub.com/#!/~MarkRizun/RewriteTool works > > Jean-Christophe >

May 7, 2015

Re: [Pharo-dev] RewriteTool

by Jean-Christophe Bach

* Sean P. DeNigris <sean(a)clipperadams.com> [06.05.2015. @12:26:49 -0700]: > Sean P. DeNigris wrote > > http://smalltalkhub.com/#!/~MarkRizun/RewriteTool/ , which is now about 70 > > commits ahead of the first! > > Alarmingly, Mark's repo seems to only be intermittently available. Many > times sthub returns 404... > > ----- > Cheers, > Sean For that, it is the trailing / => http://smalltalkhub.com/#!/~MarkRizun/RewriteTool works Jean-Christophe

May 7, 2015

Re: [Pharo-dev] NativeBoost and memory management

by phil＠highoctane.be

On Thu, May 7, 2015 at 8:53 AM, Holger Freyther <holger(a)freyther.de> wrote: > Good Morning, > > I used NB to access openlog, closelog and syslog of libc and I am not sure > if I got it right but didnât really get further by browsing > testcases/reading the > examples. > > openlog(3) takes an âchar *identâ as first parameter and the caller needs > to > make sure that the string passed remains valid until closelog is called. > > My binding looks like this: > > c_openlog: ident opt: aOpt facility: aFac > "openlog(const char *ident, int logopt, int facility);" > <primitive: #primitiveNativeCall module: #NativeBoostPlugin error: > errorCode> > ^self nbCall: #(void openlog(char *ident, int aOpt, int aFac)) > > And I call it like this: > > Identity := anIdentity asNBExternalString. > self c_openlog: Identity opt: anOption facility: aFacility. > > The things I am not sure with: > > 1.) is it right to use âchar *identâ instead of âString identâ? âStringâ > will copy > the string by value but the pointer passed to openlog would be a pointer > from > the C callstack? > asNBExternalString will do this with a NBExternalAddress: fromString: aString | result | result := NativeBoost allocate: aString size + 1. (self assert: result notNil). result writeString: aString. ^ result writeString: aString "write a null-terminated byte string to receiver's address" | str | str := aString copyWith: (Character value: 0). NativeBoost memCopy: str asByteArray to: self size: str size. So, yes, it allocates heap memory basicAllocate: numBytes "Allocate a memory block with given size in bytes, answer an NBExternalAddress instance - address to the beginning of memory block" | addr | addr := heap allocate: numBytes. addr = 0 ifTrue: [ ^ nil ]. ^ NBExternalAddress value: addr char * looks a null terminated string, which str := aString copyWith: (Character value: 0). provides So, asNBExternalString is compatible with the char * > 2.) Am I right with the call to asNBExternalString? > Should work. > > 3.) How does this integrate with GC and image quit/image resume? E.g. I > could > use >>#freeAfterUse and if I do âIdentity := nilâ at some point free > should be called > on the old string. Is there anything in the system that sets the address > to 0x0 after > the image is starting? > As the thing is a NBExternalAddress, there is a finalize self free which will do the "free" call when the Identity will be gc'd. As for the statup/shutdown, maybe startUp: resuming shutDown: quitting methods on the class side of your facility will do the necessary things. I am pretty sure the address would be pointing to invalid crap upon image restart. Doing free too many times will just raise an error I guess (which you can catch). free: address sema critical: [ | block | block := reservedBlocks removeKey: address ifAbsent: [ self error: 'Unable to find a memory block with given address' ]. block makeFreeFor: self ] HTH Phil > > thanks > holger >

May 7, 2015

NativeBoost and memory management

by Holger Freyther

Good Morning, I used NB to access openlog, closelog and syslog of libc and I am not sure if I got it right but didnât really get further by browsing testcases/reading the examples. openlog(3) takes an âchar *identâ as first parameter and the caller needs to make sure that the string passed remains valid until closelog is called. My binding looks like this: c_openlog: ident opt: aOpt facility: aFac "openlog(const char *ident, int logopt, int facility);" <primitive: #primitiveNativeCall module: #NativeBoostPlugin error: errorCode> ^self nbCall: #(void openlog(char *ident, int aOpt, int aFac)) And I call it like this: Identity := anIdentity asNBExternalString. self c_openlog: Identity opt: anOption facility: aFacility. The things I am not sure with: 1.) is it right to use âchar *identâ instead of âString identâ? âStringâ will copy the string by value but the pointer passed to openlog would be a pointer from the C callstack? 2.) Am I right with the call to asNBExternalString? 3.) How does this integrate with GC and image quit/image resume? E.g. I could use >>#freeAfterUse and if I do âIdentity := nilâ at some point free should be called on the old string. Is there anything in the system that sets the address to 0x0 after the image is starting? thanks holger

May 7, 2015