AltME groups: search

Here's a version of my last one above, but with Steeve's trick adapted 
to make a /compare option. It defaults to its old case-sensitive 
behavior.

rle: func [
	"Run length encode to series of [length value]"
	s [series!] "The series to encode"

 /into {Insert into a buffer instead (returns position after insert)}
	output [any-block!] "The output buffer (modified)"
	/compare "Comparator function for equvilance"
	comparator [any-function!]
	/local x r qr b e
] [
	unless into [output: make block! 2] x: none
	r: case [
		compare [[any [e: if (apply :comparator [:x :e/1]) skip]]]
		any-string? :s [[any x]]
		'else [qr: copy [quote 1] [(poke qr 2 :x) any qr]
	]
	parse/case :s [any [b: set x skip r e: (
		output: reduce/into [offset? :b :e :x] :output
	)]]
	either into [:output] [head :output]
]

I did choose that option when installing, but I can re-create the 
VM to try to fix a little hassle with the startup menu. What happens 
now is that on the default first option, 'Start Syllable', you get 
an error saying that the selected disk does not exist. Choosing the 
second option, 'Start Syllable (automatic boot drive search), finds 
the OS and boots. I'm not sure what went wron there - I tried to 
install grub to the MBR - basically, I pretty much accepted the defaults 
or suggestions when installing. I'll go through it again just to 
make sure that I didn't mess something up, though, since it doesn't 
take very long.

I've re-installed yet again and gotten it to work by selecting the 
first controller - for some reason the installation defaults to putting 
the HD on controller 1 as the second device - the CD being the first 
on the same controller

Does it retain the VMware settings? The problem with loading images 
from other emulators is that you don't get the emulation settings, 
so it can break when for example it defaults to SATA instead of ATA

Otherwise, it defaults to what I assume is a Unicode pair for extended 
characters.

yes... although it would probably be more "correct". then again, 
the defaults in VID also have a certain look. as long as it's simple 
to override, then it may not be a problem.

A not about the other big big bug...the recycle bug.  when I set 
my may messages to 100, I simply never see that bug.  I left my settings 
at over 1000 recently, and I knew it because I started seeing the 
recycle bug.  If this is true for others, we will change our defaults 
to start.

also not to make assumptions about what default fore/back colors 
should be.  paying attention to OS defaults can help there.

sweet, now if they were only the colors I'd want, instead of the 
white screen supremicists defaults

that's probably true.  altme stands out like a sore thumb on my screen. 
 All the other apps pretty much abide byt my defaults.

wouldn't break my heart, if AltMe used my system defaults.

sure, i agree it would be better to have user defaults.

Make sure you have AltME closed when you change it or close it then 
open it after changing it. I used "Set program access and computer 
defaults" and selected custom then picked Firefox as my default browser.

Yes, it defaults the first running world on your machine to 5400, 
and increments each server started after the first server by 1.

to me the problem is that MS should have created a real application 
framework... actually, only Apple with OSX seems to be really close 
to this.


there should be THREE root directories and only three dirs for each 
application.  windows has dozens, which is crazy.
1. Application SOURCE data (.exe, libs, datafile, etc)
2. Application STORAGE data (game saves, defaults
3. Application Temporary data, flushed on program quit by the OS.


within Storage, you'd have application and user-based storage and 
that's it.


The os refuses to execute any file not within application source 
dir, so you have already clamped down on a lot of security issues. 
 a list of approved .exe is stored whenever you launch an application 
for the first time.


for a bit more flexibility, several application dirs could exist, 
so that you can do stuff like compile items and run them locally 
within your dev environment, but these dirs would be explicitely 
registered within the os.


on install the Application should be required to ask permission (like 
flash) in how much space will be stored within STORAGE if it exceeds 
a max default, independently of SOURCE.


The implementation of MyXxxx folders is completely horrendous, and 
should be a simple and ubiquitous form that has one entry for each 
media type, and a browser where you want to put your stuff (which 
might not be user-specific).


IIRC this was addressed in Vista, but then they went and added aliases 
to everything, which raises errors in the explorer, cause you can't 
use them.  They alias core windows directories in other languages, 
the result being that explorer and actual disk info are different! 
 this completely fucks up many applications (like rebol).


windows has a hard time "fixing" things... they always just patch 
it... its worse since they have a hard time starting with any good 
implementation of anything to begin with.

I used it primarily for data, to add default fields. 
 item: make context[ defaults ] load %item.r

there are no functions, so no problem. now thats construct/with, 
even no binding nowhere :)
But lately i did something obscure and it drove me made..

here is a switch that defaults to 'all

Very, considering 0:0 defaults to hours:minutes -
>> 0:0
== 0:00
>> 0:0.1
== 0:00:00.1

Quick question .. how do you set the century to-date defaults to 
?

>> to-date "1/10/40"
== 1-Oct-2040

is not what I want.

I think it is importan if you get dates as strings, you dont know 
after the converion if it was 2 or 4 digits. makes the smartness 
of to-date useless, going back to self-made conversion. OTOH in most 
cases defaults are enough, but so are english month :)

So they are just handy defaults. I think they're as well-named as 
they can be, probably.

write converts linefeeds to OS defaults unless you use /binary

a lot of the APIs have changed in Windows vista/7 (not sure about 
Clipboard).  

the kind of changes I would expect...  if its not using the latest 
apis, newer features in the OS could simply be preventing REBOL from 
accessing this device.

just like how these new OS prevent you from writing within the program 
files folders.


some things which had defaults, maybe now require explicit flags... 
etc...

Just my 1 cent contribution: there is a bug with the drop-down face 
when no action is set. The 'action facet defaults wrongly to 'show-dropdown, 
making the drop-down to freeze when the user does not override it. 
The fix is simply drop-down with [action: none]

I think thare are two slightly different goals. One is to deal with 
the main data in a face. That should be a single value. The other 
is to make a snapshot of a layout, save it, qui script, and reload 
it later. I like that behavior. And i am not sure a single value 
is sufficient for that. But it would be an internal thing about saving, 
as api i prefer something with more defaults, ala VID.

I'm just going with the ugly defaults most of the time. :)

thanks for the heads up... If you are going to rewrite it... please 
allow more selections in this kind of thing... curved interpolation 
IS much better, for things like shadows, for example... but the need 
to select filter types (from all available... not just one or two), 
interpolation curve types, start-end snap on beziers, things like 
this are all essential, if we want to allow the next REBOL to be 
able to properly compete.  


defaults are good and often this is all that is needed, but for any 
graphic app, this quickly becomes essential (as you will agree, on 
your own usage), and is a weekness of current toolset.

Careful with SQLite and big blobs.   A blob can be SQLITE_MAX_LENGTH 
(usually a billion) but SQLITE_MAX_SQL_LENGTH defaults to a million. 
 A single line of SQLite may limit the size of blobs to under 1 million.
See http://www.sqlite.org/limits.htmlfor deets.

Sorry, defaults for what ?
The above changes allow eg:
	layout-pdf [ [ image logo.gif ] ]
	layout-pdf [ [ image 20x150 200x96 logo.gif ] ]

there's also to say, that the layout-pdf dialect is still meant to 
be rather low level. a human would write in QML or makedoc or some 
high level dialect, not layout-pdf; thus the defaults for image would 
be useless (even if a human was to write it, they'd only cover some 
10-20% of the cases, since not all images have the same dpi, and 
99% of the time you don't want the image at the bottom left corner 
of the page).

Graham;  Complete aside...and very dependant on firewall setiings. 
 (I have a firewall/router so the local unix box is not double firewalled...yet) 
but this lets you run Cheyenne from user process.

#!/bin/bash
#

# This script is to be run from root to allow Cheyenne to be run 
as a user
#

# NOTE: dev is eth0, spike and chester are eth1 (eth0 for the firewire)

#       this script defaults to eth1, so no longer functions on dev
#
# redirect external incoming port 80 to 8080

iptables -t nat -A PREROUTING -p tcp --dport 80 -i eth1 -j REDIRECT 
--to-port 8080

# redirect localhost port 80 to 8080

iptables -t nat -A OUTPUT -p tcp -d 127.0.0.1 --dport 80 -j REDIRECT 
--to-port 8080

I used YaST on Suse to setup Samba and accepted all the defaults. 
 I see groups, profiles, and users .. but when I try to login, I 
can't authenticate.

the ubuntu was is gnome based and uses primarily gnome based software, 
while kubuntu uses kde as the desktop manager and preferences kde 
based software.   the applications can be installed on either, but 
these are the defaults.  i personally use xubuntu which uses the 
lighter weight xfce desktop manager and installs less applications 
by default.  but all the systems use the same base libraries.

DanielSz;  Depending on how many specific associations you have created... 
temporarily moving (ie backup) the  ~/.nautilus subdir may help. 
It holds a metadata subdir.  Which if you move .nautilus Nautilius 
will revert to defaults.  


Or check out these intstructions on update-mime-types.  It effects 
~/.local/share/mime/packages/Override.xml
https://answers.launchpad.net/ubuntu/+question/6795

Does CentOS have update-rc  by any chance?   update-rc.d httpd defaults 
 will configure the init.d scripts for all default run levels on 
Debian.   For CentOS (I don't know CentOS utility bits) try
# chkconfig --level 235 httpd on
# chkconfig --list httpd

Which will set httpd to run at levels 2 3 and 5.   *Assuming # is 
the superuser prompt, not the bash comment.

Not sure there is a good solution other than to require optional 
parameters to have keywords to identify them.

Then you need to take only one pass at the input, and use defaults 
(or ignore) parameters that are not present.  At the end of the pass, 
you can then see if the collection you have makes sense (maybe some 
optional parameters are needed if others are also present).

Basically, its raw data validation....Boring, but necessary.

i (if-true-mu) and e (either-true-mu), which can be remapped but 
are by default mapped to their equivalents "IT" and "ET" (there's 
also WT for while-true) are pretty good one-character ones, but even 
those can be questionable.  But they are reasonable defaults.  The 
point of one-character-choices in Rebmu is to anticipate the need 
for flexibility for redefining them, to use as variables or whatever.

Net is going to need restricted, I think, and cache probably needs 
some restriction as well.  I wonder if there's been any discussions 
on how to keep rebol from being the target for some very malicious 
code.  I know more about IE's security model and what I can do at 
IE defaults, and would like to code such that I am not relying on 
features that can/should/will be restricted by default.

http://www.compkarori.com/reb/download-manager.r


is a wrapper around Anton's batch download script.  It reads a configuration 
file(s) from a site and loads the data into the script.

This is as: file description url and then extended file description.

Defaults are embedded into the script and are used if the read fails.
You can use it to download iso files etc.

Shouldn't that be part of show-text ?  If you're going to preset 
a number of fields eg. to reset some defaults, you don't really want 
to rebfocus face on each one as well as do a show-text ?

Robert: "Is there a way to re-use widgets like done with 'style in 
VID? For example: I have a drop-list specification, I need at several 
places but I don't want to copy the code."

Short answer, no.


If you have a "style" that is sufficiently different from the base 
widget (that is used often) then consider:

1) modify the widget's defaults in %rebgui-widgets.r, or
2) modify them inline, as in:

	ctx-rebgui/widgets/<widget>/<facet>: <value>

3) derive a new widget that works exactly the way you want.

Robert:
	1) global words: they are not. Try the following:


  f: has [v] [display "Test" [v: field button [unview halt]] do-events]


 My normal approach is to wrap each display in its own function with 
 something like:


  display-account-summary: make function! [id [integer!] /local f1 
  f2 f3 f4 f5 f6] [
			...
			display "Account" [
				f1: field
				f2: field
				...
			]
			...
		]


 2) tab-panel: works, just take a look at how it is used in %tour.r. 
 Note that it defaults to "size: -1x-1". Maybe you need to make the 
 panel contents bigger?

No, type conversion is wrapped within an attempt and defaults to 
previous value on failure.

It changed slightly with Cyphre's changes in that not specifying 
a number (or none) now defaults to selecting the first item. If you 
want to start with no items selected then you must use none (0 does 
the same thing, but causes problems for picked and selected until 
a "valid" item is selected ... I'll fix that for the next build).

build#60 committed to SVN. Added Henrik's button widget with 2 minor 
modifications:

	1) Over color defaults to colors/over
	2) All of init inlined into the effect facet


The first change is self-explanatory, the second follows the principle 
that init should do as little as possible (facet code is evaluated 
once while init code is evaluated every time the widget is used). 
This change has one subtle side-effect, the "Refresh Display" button 
of %tour.r no longer works for all widgets (button in particular). 
This will be fixed in a future build.


One thing to note about the new button widget is its default size: 
15x6 instead of 15x5 units. This should not be a problem for most 
buttons, but may have spacing/alignment issues for inline buttons.


Note that the button change necessitated a small change to request-date 
which is now working again.

Oh, your defaults.  monspace and serif, everything but courier on 
this box.

Uploaded build#81 to SVN. Includes the following changes:

	Fixed label & text auto-sizing (now defaults to -1x-1)
	Spinner now accepts an empty starting value
	hilight-on-focus now works again

 [edit-list field password spinner] are now hilight-on-focus (area 
 remains caret-on-focus)
	edit-list fix

Also updated:
	http://www.dobeash.com/RebGUI/widgets.html
	http://www.dobeash.com/RebGUI/functions.html

I think if you use t/redraw it always defaults to no sort

 Is there any way to circumvent this for items that need a specific 
 pixel size, in my case a picture?

 image defaults to size -1x-1 which means *not* specifying a size 
 will default it to the image size.


 Is it possible to resize the window when my picture is changed, so 
 that the new picture doesn't get resized to the previous one's size?
 This should get you going:

	display "" [
		button [face/parent-face/size: 320x240 show face/parent-face]
	]


Note that in this case the size *is* in pixels as you are modifying 
a face object directly (as opposed to specifying a wudget's *unit* 
size). Hope that helps.

Should be both, no?  Has the ability to remember, but defaults to 
what most people expect, with the ability to override.

I don't like defaults because you don't know if the user made that 
selection, or forgot to make one.

Use set-values ;)


The problem with clear is that it is context dependant, and could 
be:

	""
	0
	none
	true
	false


quite often the definition of 'clear-values would be more akin to 
a reset function; "restore these widgets to my app-defined set of 
defaults" ... far better to make it explicit with 'set-values then.

Defaults to auto now, and has for some time.

Out of interest, the memory figures on OS/X using rebface are:

	rebface	1.8Mb
	VID		3.2Mb after loading %view.r
	RebGUI	3.5Mb after loading %rebgui.r


but these are for the respective defaults (VID apps often require 
additional patches/fixes and styles, RebGUI apps rarely use *all* 
the default widgets). I should also add that RebGUI (or indeed VID) 
have ever been fully profiled and optimized.

I just used the tour.r on the latest rebgui.

There you can select a color for a button, but as soon as the mouse 
hovers over the button, the colors switches to the defaults.

Is this by intent or is there a way to keep the selected color(s) 
?

Build 205
- Added request-list
- Added request-email
- Added do-events
- Updated request-spellcheck (added image)
- Updated request-error
- Updated request-value
- Label size now defaults to 25x5 (was -1x5)
- Replaced "attempt" logic tests with "error? try"
- Memory footprint drastically reduced
- Added no-wait to display func
- Updated request-progress to use display/dialog/no-wait
- Fixed button over state bug
- Fixed slider jerkiness

Note the reworked tree widget only has one option [only], and currently 
defaults to expanded mode with no leading graphics. It accepts input 
in either nested block or file/path notation. Selected items are 
now fully qualified by default. Due to it's use of face-iterator, 
it no longer has practical data limits (previous version created 
a face for each item).

Build 214
- updated toggle (2x2 edge on select)
- added missing btn widget
- added missing chat widget
- added flash requestor
- added request-about
- renamed request-password to request-pass

Flash is used as follows:

	...
	flash "message"
	...
	hide-popup
	wait []
	...

request-about is used as follows:


 request-about/url "MyProj" 0.0.1 "Copyright (c) MyCo" "www.site.com" 
 ; or a url!


Note the logo is located in ctx-rebgui/images/logo and defaults to 
the info icon ... replace with your own logo image as follows:

	ctx-rebgui/images/logo: load %my-logo.png

Really? I went positional: ["element" "namespace" ["attribute" "value"] 
["subelement" ...] "text" ...]

with missing namespace or attribute block being #[none], so defaults 
can be done with ANY.

The inconsistency between R2 and R3 is because R3's reflection model 
was changed to be more secure, and easier to sandbox. In R3 you can't 
get a reference to the real spec and body of a function after the 
function is created, at least from non-native, non-debug code. That 
is why the hack above required saving a reference to the spec from 
before the function was created; if you don't do that, you won't 
be able to get at the spec or body afterwards if your security settings 
are set to the defaults (and turned on - that's another story).

So one thing I would like to suggest is that the "Change" button 
in the installer defaults to the folder as indicated in the default 
path provided when you start the installer.  I think this is the 
root of the problem as the text looked like it was fully provided 
but the path field was short and we didn't confirm the full path. 
 In my opinion it should have included \Rebol\View, but believe it 
did not.

clear buffer

 is also an optimization as it defaults to 32K values (make block! 
 1032 * 32) and I don't won't to reallocate it each SQL call. The 
 following benchmarks (Transactions Per Second) give some backround 
 to the design decisions I made:

	buffer				1744718
	copy buffer			282
	copy/deep buffer		76

	clear buffer			1144733
	buffer: copy []			824352
	buffer: make block! 32768	387


So the approach I took optimizes for large result sets by allocating 
a large buffer once up-front and then just referencing it thereafter.

/no-copy is a reasonable compromise, and fits with the /flat refinement 
in that the driver defaults to the expected "safe" behaviour. I'll 
add that to the next build.


XML output: I'm adding HTML output to the next build; is XML needed 
in addition to that? And if so, what does a "standard" table definition 
look like under XML (please provide a short example).


SQL refinements: the idea is that in most cases you specify all the 
refinements you need with CONNECT and just use SQL without any; except 
in the following two special cases:


1. SQL/direct: where you used CONNECT (without the /direct refinement), 
but need to access a *specific* table that does not need its TEXT 
column values MOLDed (e.g. sqlite_master)


2. SQL/raw: as above but flattens data. Used primary to access system 
tables / data, this ensures that values are always returned in a 
consistent format; lets you write code like:

	to file! third DATABASE


without having to worry about the format of the returned block(s).


Grammatical question. Do folks prefer I use the word INDEXES or INDICES 
to refer to more than one INDEX?

The log-path issue is best resolved by adding a log-file word to 
the sqlite context that defaults to %sqlite.log. You can then do 
the following in your code:

	sqlite/log-file: %my-path/my-log-file.log

but... on-get-value has nothing to do with "defaults"?

REBOL [
    Title: "Formatting dialect"
    Author: "Carl Sassenrath"
    Date: 01-Jun-2007
    Version: 2.99
    File: %format.r

    Purpose: "Simple, powerful string formatting; well padding at least"
    History: [
        01-Jul-2007 2.99 "Carl" "first R3 alpha"
        02-Sep-2007 2.99 "Bluey" "prepped for R2"
    ]

    Comments: {R3 format will be more powerful, this is a teaser version}
    Usage: {
        format rules values
          /pad padding-character; defaults to space
        rules include the following dialect messages;

          integer!              sets width left aligned (trailing spaces)

          negative integer!     sets width right aligned (leading spaces)

          string!           inserts the string based on position in rules

          char!          inserts the character based on position in rules

        rules can be a single message or a block! of formatting messages;

        rules can also be a word evaluating to a single or message block!
        Examples:
          format 5 12 == "12   "
          format -5 12 == "   12"

          format/pad [5 "---" -5] [12345678 1] "0" == "12345---00001]
    }
]

rxvt quick tests;  stdio is not reliable (with the defaults - these 
are quick tests)  

$ ./rebhost.exe -q   >> call "ls"    works    >> call "less readme.html" 
  does not detect console.
$./rebhost.exe < small.r  does not work

so in short, not a good choice for redirection, but seems ok for 
the testing I've been doing.  Haven't tried running gnu readline 
with it yet to see if command line recall will work.  I've gotten 
too used to editing source and do'ing.

In C++ for example the defaults are set as follows:

int fnc (int x = 1 int y = 2){...}

I can understand the syntax issue from the perspective of it requiring 
the parameters defaults be ordered.

Anton, we are trying to come up with sensible defaults for all of 
the options. Get in on the conversation in R3 chat :)

What about using a word which requires three paramaters the first 
and third of which must be those delimiters ( smart might be to have 
3 sets of delimiters which need not even be paired - or to have what 
counts as a delimiter set for the user context with one pair of system 
defaults.

possibly the windows version defaults to 16 bits more quickly than 
linux and OSX versions...  :-/

so 'parent is not set at all? That is bad then. So you have to preset 
your parse level words to some defaults, if you can't be sure, if 
they will be set?

Yeah, in QML we ran into the same font problem.  So what we did was 
make Times, Helv, and Courier forced to be the defaults, then allowed 
any font name to be called as bassed to a variable as the real name 
of the font.  On windows you can have font names with spaces "Times 
New Roman" for example.  By have the top three just be one simple 
word everyone can remember I figured it would make people happy.

I was relying on info from http://en.wikipedia.org/wiki/Greylisting
about other MTA default delay :


Most open-source MTAs have retry rules set to attempt delivery after 
around fifteen minutes (Sendmail default is 0, 15, ..., Exim default 
is 0, 15, ..., Postfix default is 0, 16.6, ..., Qmail default is 
0, 6:40, 26:40, ..., Courier default is 0, 5, 10, 15, 30, 35, 40, 
70, 75, 80,...). Microsoft Exchange defaults to 0, 1, 2, 22, 42, 
62 ...

Just using the defaults with absolutely no customizations!

it defaults to port 80 ...

I think that the Accept, Accept-Charset and User-Agent headers are 
the defaults, and spec/headers are the user-specifiable options.

but using a timeout for some reason means that it defaults to the 
timeout value ?

system/dialects/text is handy. Seems the shortcut b, i, u and nl 
words are not yet implemented/working. 'drop is cute, and I note 
that 'drop without an argument defaults to drop 1.

ok, try 'defaults

Right, I didn't realize at first that you were talking about facets. 
I thought you were talking about faced. Defaults are what the facets 
object in the style object is. The faced objects are the prototype 
of the specifics (non-defaults).

direction would be a tri-state 'up 'down none.


none means inherit.   all gob types would have their defaults set 
to one of the three states.

The problem is we are trying to eliminate the need for using column 
definitions in case of one row/one column layout.

Currently you would write:
view [
	panel 1 [
		panel [
			button
			button
		]
		panel 1 [
			button
			button
		]
	]
]

or in case of 'inverted defaults' for panel without any number:

view [
	panel [
		panel 2 [
			button
			button
		]
		panel [
			button
			button
		]
	]
]




Which would make two panels in one column. First panel would have 
2 buttons in one row and second panel 2 buttons in one column.


I personally think this description of the same layout would be more 
readable:

view [
	vpanel [
		hpanel [
			button
			button
		]
		vpanel [
			button
			button
		]
	]
]


So the key is to improve readablility of the layout definition in 
cases that will be used from 99% IMO.

Anyone have better solution?

Back to ACROSS and BELOW, then? Those have never been misleading, 
they just weren't styles in R2/VID but layout directives which were 
lost after the layout phase finished. And make the difference between 
PANEL, GROUP, TIGHT behaviour an attribute to them, with reasonable 
defaults. Would that even work?

layout [
	below [
		across [ ... ]
		below [ ... ]
		below 'tight [		; TIGHT behaviour
			button
			button
		]
	]
	across [
		...
	]
]

Robert - agreed - as much as there is a way to override some "defaults" 
by the stuff I want, I am mostly OK. Of course I still care for things 
being easily useable, if possible ...

no. having an engine which provides great GUI defaults is essential, 
but not at the cost of being able to tweak a widget .


skins/themes/stylesheets provide usefull defaults,  but having access 
to overide any of this is absolutely necessary.

but when a client tells me, I want this banner red, this one navy 
and this one black... I've stopped trying to convince them that its 
ugly, it just irritates them, and it inevitably leads to bad relations. 
   I will convey my experience and state that its not something professional, 
but in the end, the client writes the check, and I need to be able 
to push the bytes out the door.


there is no philosophy or ideology when you need to deliver and a 
tool can't turn around and be flexible.


I don't want to post stuff from other engines here since its not 
a comparison game, but I've used many APIs from prbably 20 different 
dev platforms, and everytime I use one which has an "unwielding" 
ideology where you can't modify things to make them do what you want... 
as a user, I get frustrated and I just look for something else to 
do and/or work on.   


good defaults, decent properties and backbone, clean style.  all 
the rest, open and hack.  I woudn't be a Reboler otherwise.

that's just my 2 cents.

There are real learning and semantic advantages to just going with 
one return model. We just need to make the limitations of whatever 
model we choose easy for regular programmers to workaround if necessary, 
and pick the defaults well so the workarounds won't need to be specified 
as often. The last model satisfies all of those at the expense of 
losing the benefits of dynamic return, and the next to last doesn't 
even lose those, though it does lose some simplicity. Given that 
the remaining benefits of dynamic return can be restored by keeping 
THROW dynamic and fixing the THROW/name bugs, I'm willing to part 
with dynamic return and get back the simplicity.

The important issue is that the defaults match, if we want to have 
any.

Another important issue is that the defaults be settable on a per-platform 
or per-app basis, and that the behavior of Red/System is explicitly 
defined in a cross-platform way. We don't want to repeat the mistakes 
of C just because we're interfacing with C.

We'll probably still want to cater to the vast majority of C interfaces 
which just uses platform/compiler defaults.

Right, we need settable defaults that can match the local C standard 
or whatever it is that sets the standard locally (JNI perhaps).

C structs are what I'm talking about, particularly ones declared 
with the pragmas. The quirks can go in the settable defaults, including 
things like the size of certain datatypes that have ill-specified 
types in the C standard.

You are a little spoiled on Unix-like platforms, Andreas. On Windows 
there is no standard C compiler; there are a bunch of competing ones 
instead. Each has its own quirks and defaults. Since Red is not itself 
written in C, you can't just go with the quirks of the same compiler 
it is written in. And different libraries are compiled by different 
people with different compilers. If you want to interface with them, 
you have to be specific.

I found the same info that Windows system libraries use stdcall, 
but that MSVC defaults to cdecl. I had been compiling the binding 
with stdcall, and both work. I standardised on cdecl and retained 
the comment for the moment being

The obvious defaults and accessors that you would expect often aren't 
there

For the floating point format experts around: 


I am investigating the simplest and most accurate way to support 
floating point comparison by implementing an `almost-equal` operator 
in Red/System. Due to the intricacies of such task, I have selected 
the two algorithms that look the best to me:


1) Tolerance-based: http://realtimecollisiondetection.net/blog/?p=89

2) Ulp-based: http://www.cygnus-software.com/papers/comparingfloats/Comparingfloating point numbers.htm


Of course, in both approaches, there are some accuracy parameters 
to be set by the user to adjust the comparison formula (good defaults 
will be provided anyway).


Are there better algorithms or better implementations of methods 
1) or 2)?