comp.lang.smalltalk Google Group

Swazoo 2.2 beta with fast file upload

2008-07-26T17:53:10Z

Dear Smalltalkers,
New Swazoo 2.2 is ready for beta testing, with much improved upload
(input streaming) performance as a main new feature.
On Squeak it uploads 15 times faster than before, achieving 1.5MB/s
throughput locally on my Linux 3.2GHz PC.
On VisualWorks it is even more impressive: 30 times better, achieving

Re: The Hitch Hiker's Guide to the Smalltalk Compiler

2008-07-24T16:17:21Z

In article <EGOhk.16223$uE5.10...@flpi144 .ffdc.sbc.com>, Eliot Miranda says...
It also appears to be hosted at
[link]

Re: The Hitch Hiker's Guide to the Smalltalk Compiler

2008-07-24T16:17:22Z

In article <EGOhk.16223$uE5.10...@flpi144 .ffdc.sbc.com>, Eliot Miranda says...
It also appears to be hosted at
[link]

Re: The Hitch Hiker's Guide to the Smalltalk Compiler

2008-07-23T22:56:45Z

No worries; Vassili remembered web.archive.org and we have it now.

The Hitch Hiker's Guide to the Smalltalk Compiler

2008-07-23T21:58:27Z

Hi All,
anyone have a copy of this they can let me and/or Vassili have? It
used to be in smalltalkchronicles.net but that's defunct and Vassili
might have a copy on a CD but not online. If you have a copy please
email it to Vassili and/or me
TIA

Re: turn of the Cog

2008-07-23T20:35:25Z

Its a lot simpler than some other Smalltalk code I could mention
(Croquet for example is complicated!). Its only the VM; simple
mechanical stuff :)
Well I made sure the decompiler works. But what doesn't work is the
feature that let's the decompiler use temporary names. he issue here is
that the existing scheme assumes there is just one vector of temp names.

Re: The Builder Pattern to rule them all Re: Smalltalk syntax can rule them all

2008-07-23T19:10:21Z

Ed> How are nested elements handled?
Ed> <body>...<OL>...<LI>item <EM>one</EM>...</OL>...</body>
html orderedList: [
html listItem: [
html text: 'item '.
html emphasis: 'one'.
].
].

Re: turn of the Cog

2008-07-23T19:18:19Z

Eliot Miranda a écrit :
Great explanations.
You make it look simple.
The bootstrap
I understand you'd love to be helped with the Decompiler, every modest
change I attempted about bytecode generation like inlining #repeat loops
would already be integrated without this tricky beast.
Nicolas

Re: The Builder Pattern to rule them all Re: Smalltalk syntax can rule them all

2008-07-23T13:48:57Z

How are nested elements handled?
<body>...<OL>...<LI>item <EM>one</EM>...</OL>...</body>
ed

turn of the Cog

2008-07-23T01:16:34Z

Hi All,
I've made a significant step today in making available the first
cut of my closure scheme. There is a new blog post on
[link] about the closure bytecodes which at the
end includes a pointer to the bootstrap @
[link]. The bootstrap

Re: to:by:do: optimization and trivial Shootout benchmark improvment

2008-07-22T20:28:50Z

>>to: stop by: step do: a1block
| each b s e |
each, b := self.
e := stop.
(s := step) positive
ifFalse: [b := 0 - b. e := 0 - e. s := 0 - step].
b > e ifTrue: [^nil].
b := e - b.
[ a1block value: each.
each := each + step.
(b := b - s) positive
] whileTrue
--cstb

Smalltalk-vn : a new mailing-list for Smalltalk in Vietnam

2008-07-21T21:48:04Z

There is a now a Smalltalk mailing-list for Vietnamese users or
developers. You can subscribe here :
[link]

Re: to:by:do: optimization and trivial Shootout benchmark improvment

2008-07-21T21:41:04Z

paolo.bonz...@gmail.com a écrit :
Yes, current implementation is inlined (for positive step):
element := start.
[element <= stop]
whileTrue: [aBlock value: element. "inline this block"
element := element + step]
Replace <= with >= in case of negative step.
One alternative way to write the loop is like implementation of

Re: to:by:do: optimization and trivial Shootout benchmark improvment

2008-07-21T06:33:25Z

Yes, it does. The problem is that you need to use #>= or #<= to do
the test, depending on the sign of the increment.
Another alternative, if you have from/to/by on the stack, is to change
the upper bound to something like "to + by - ((to - from) \\ by)", so
that you can use #= as the condition for exiting the loop.

Re: The Builder Pattern to rule them all Re: Smalltalk syntax can rule them all

2008-07-21T06:20:39Z

this is more or less how you describe pages in seaside, actually.
Paolo