Is new Gemstone FFI non-blocking?

Mariano,

I'm pretty sure that the FFI calls in GemStone 3.0 (with the new FFI interface) are all blocking calls and the whole vm is blocked. I'll double check this morning when I get into the office ...

Dale

----- Original Message -----
| From: "Mariano Martinez Peck" <[hidden email]>
| To: "GemStone Seaside beta discussion" <[hidden email]>
| Sent: Thursday, April 4, 2013 7:40:10 AM
| Subject: [GS/SS Beta] Is new Gemstone FFI non-blocking?
|
| Hi,
|
| While I was thinking about a possible port of DBXTalk to Gemstone, I
| wondered with the new FFI implemented in Gemstone is non-blocking. Second,
| which is the granularity? All block/all non-blocking?  or in each mapping
| (method - C function) I can chose to be sync or async (as it is in VW).
|
| Thanks,
|
| --
| Mariano
| http://marianopeck.wordpress.com
|

----- Original Message -----
| From: "Mariano Martinez Peck" <[hidden email]>
| To: "GemStone Seaside beta discussion" <[hidden email]>
| Sent: Thursday, April 4, 2013 7:52:51 AM
| Subject: Re: [GS/SS Beta] Is new Gemstone FFI non-blocking?
|
| On Thu, Apr 4, 2013 at 11:45 AM, Dale Henrichs <[hidden email]> wrote:
|
| > Mariano,
| >
| > I'm pretty sure that the FFI calls in GemStone 3.0 (with the new FFI
| > interface) are all blocking calls and the whole vm is blocked.
|
|
| Uhh what a bad news :(  Being Gemstone and its philosophy I was expecting a
| non-blocking (multi-threaded FFI)  :(

I still might be mistaken on this point:) I'll have definitive answer when I get into the office.

|
| I'll double check this morning when I get into the office ...
| >
| >
| Thanks. If it blocks the whole VM it means the current Gem? what happens if
| I have more gems?

Only the process/gem making the call would be blocked by an ffi call ... we achieve most of our our concurrency using multiple (independent) gems rather than running multiple threads within a gem ...

Dale

----- Original Message -----
| From: "Dale Henrichs" <[hidden email]>
| To: "GemStone Seaside beta discussion" <[hidden email]>
| Sent: Thursday, April 4, 2013 8:16:38 AM
| Subject: Re: [GS/SS Beta] Is new Gemstone FFI non-blocking?
|
|
|
| ----- Original Message -----
| | From: "Mariano Martinez Peck" <[hidden email]>
| | To: "GemStone Seaside beta discussion" <[hidden email]>
| | Sent: Thursday, April 4, 2013 7:52:51 AM
| | Subject: Re: [GS/SS Beta] Is new Gemstone FFI non-blocking?
| |
| | On Thu, Apr 4, 2013 at 11:45 AM, Dale Henrichs <[hidden email]> wrote:
| |
| | > Mariano,
| | >
| | > I'm pretty sure that the FFI calls in GemStone 3.0 (with the new FFI
| | > interface) are all blocking calls and the whole vm is blocked.
| |
| |
| | Uhh what a bad news :(  Being Gemstone and its philosophy I was expecting a
| | non-blocking (multi-threaded FFI)  :(
|
| I still might be mistaken on this point:) I'll have definitive answer when I
| get into the office.

I remembered correctly ... the FFI calls are blocking calls, so the pattern one would want to follow when consider using long running FFI calls is to isolate those long running calls in a separate vm similar to the approach that we're using with the Service VM[1] ...

Dale

[1] http://code.google.com/p/glassdb/wiki/ServiceVMExample