> On 29-08-2016, at 7:45 AM, Bert Freudenberg <bert at freudenbergs.de> wrote:
> 
> IMHO we should reconsider how project saving works. ImageSegments were a good technology back when machines and the interpreter VM was slow. Now we have faster machines and VMs so this very VM-dependent mechanism is no needed anymore.

Although I would have to comment that  there are still (relatively) slow machines still in major use, in general I agree that we could almost certainly do without ImageSegments.

> 
> Yoshiki's S-Expression format is one way to serialize, yes. We have used it for the QuickGuide projects in the Squeakland image. In the short term it's a good option. However, for more general project saving we may want to take a look at other serialization schemes.
>  
> I think we should deprecate ImageSegment. Is there any active use of them? I seem to remember SqueakMap used them but I'm not sure.

The last use *I* had for them was for exobox (so…. 1999? Were we partying? Ah, no, my archives say 2000 so we had given up on raspberry berets and indeed were shortly to give up on exobox) to load fonts very fast. It was indeed *very* fast compared to reading them any other way. However, I also have an email from Andreas of 5sep02 where we were discussing ‘script squeak’ and he asserts that some measurements convinced him that what he described as a ‘bytecode loader’ was faster, in part because of no pre & post ImSeg messing around.

I have no idea if this is a profound thought or a trivia that everybody knows, but the deep-copy code seems awfully like (much of) a serialiser


tim
--
tim Rowledge; tim at rowledge.org; http://www.rowledge.org/tim
Useful Latin Phrases:- Noli me vocare, ego te vocabo = Don't call me, I'll call you.