AltME groups: search

eFish... 1.2.40, also tried latest beta 1.2.57, and went back to 
1.2.8 to see if it handled it any different.

Tom: Yes, I was aware of read/lines and how it is similar (apparently 
not the same) as parse/all series "^/".  read/lines worked just fine.

I don't know why last night I wasn't happy with read/lines - must 
have been tired!

and what do you do if the text is not a file ?  Write your own parse 
rule.

How do you break a parse rule and still have it be true ?

Actually, what I was wondering was how to break out of an action 
in a rule and still let it return true

I'm using these rules in my server side implemention to the top command


one-line-rule: [copy line thru {^/} ( if line = ".^/" [ line: join 
"." line ] write-client line)]

header-rule: [copy header thru {^/^/} (write-client header write-client 
)]

msg is the email message including header and body
lines is the number of lines requested by the TOP command

parse msg compose [ header-rule (lines) one-line-rule ]


Now, I can't check the parse syntax as rebol.com is down, but I seem 
to always get the whole email with my header-rule and not just the 
header.

Graham,  I'd probably use parse/all rather than parse. Also don't 
forget the parse-header function and all the associated bug fixing 
work related to it in view 1.3 project. May or may not be of use 
to you.

It wouldn't have to be industrial-strength, but it would like a security 
blanket for developers experimenting with the new language. PCRE 
is found all over the place in languages on Linux machines, and the 
absence makes some developers uncomfortable - despite the fact that 
Parse is better.

yea but then rebol programs would start getting comtaminated with 
unfriendly gobbeldy gook and rebol developers would have to learn 
pcre

What about a parse rule that takes pcre strings as input and produces 
a parse rule as output?

I am not totatly against REs I use them all the time in shells, and 
having them built in would make writing "work alike" programs easier 
but over all , it  seems to me like a step down

I will be a few to make concreat but basicly you work with what is 
common to all lines , in this case colons and newlines

so once you have done some header-lines  and got the ones you are 
interested in you skip the rest with thru "^/^/"

actually, I copy the header  and body out first and process them 
separately.

and if you had a header with a line that did not begin with  'Date, 
From, Subject or To  then you could prematurely break out of header-rule 
 before you got all your bits

While we're here .. what this taint thing that Perl has, and is it 
a concern for Rebol ?

tainting forces you to consider the users input  and explicitly allow 
it to pass

Graham: for your header, like Brett said, parse/all is needed when 
you work on strings with newlines and spaces. last line should be:
parse/all header [header-rule some [ thru "^/" header-rule]]

and being very careful to never  effectivly 

do [ user-input]


without being sure user-input  could not cause unintended side effectd

I like CamelCase words. Simple to remember and use. IIRC camelCAse 
is not a valid CamelCase word. But anyway, it depends how I teach 
my users :-))

http://en.wikipedia.org/wiki/CamelCase... CamelCase is referred 
to UpperCamelCase, and camelCase is referred to as lowerCamelCase

Tom, your example doesn't terminate, like mine. The thing IMO is 
that the last Word is a CamelCase word and the 'end condition is 
somehow missed. It nevery reaches the halt.

Robet you also have to worry about  YaBaDaBaDoCamelCases  (even and 
odd) 

to get it to return true ,  figure out what is left when the outter 
most  some finishes.
parse ...[
	some [
		...
	]
	copy remenant to end ( print remenant)
]


then make  the  your rule cpnsume the remenant ok if you  don't care 
just put a  
	to end
there

You can either put your parse in a catch [] and throw a true if not 
flowtext 
or something like this
parse/all/case test-text [
	some [

   copy camelcase-word [upper-case some rest-chars upper-case any rest-chars] 
   (
		 	if not empty? text [?? text clear text]
		 	print ["CamelCase word found:" camelcase-word]
		)

  | copy flowtext [some [rest-chars | upper-case] any delimiters] (
			append text flowtext
		)
		| copy flowtext [some delimiters] (
			append text flowtext
		)
	] to end
]

what you could do, is extract Didier's implementation of the TOP 
command, and then get the first line of each header in your mailbox. 
 If it has the return-path set to <>, then note it in a list.  When 
finished, go thru and issue deletes on all of those.

Hmm... my parse still not termines the 'some part. I never reach 
the end. The problem is that the rest of the string is "" and this 
seems not to be handled.

newby here: can anyone direct me to a sample of  code which matches 
a pattern over multiple text lines? I need to process a 5MB text 
file and remove all patterns of multiple consecutive email address 
es eg. [foo-:-my-:-com]; [foou-:-you-:-net] except the multiple email address 
string spans mulitple lines. Thanks for any pointers

and the multiple email address string occurs multiple times ;)

Michael, this is going to be a very general response to your request. 
 Review setting up parse rules and use something like...
   parse text any [rule1 | rule2 | rule3]

Here's some Parse documentation from the Rebol/Core guide to get 
you started.  Share your progress and questions, maybe a line of 
sample data or two and maybe I can be of more help.

Sometimes it helps to parse in two steps. a loop for each line-group 
and parsing that group seperately. becaus ethen 'to/'thru work better.

Can somebody explain me why 'parse fails in the first case and returns 
true in the second case ?

r: [any into ['a (print 'a)]]
t: [[a][a][a]]
print parse t r 
->
a
a
a
false

r: [any [into ['a (print 'a)]]]
print parse t r 
->
a
a
a
true 


In the second 'r(ule) the additional [ ] make it kind of explicit, 
but shouldn't return the first version true as well ? Am I forgeting 
something what 'parse "thinks" when looking at the first 'r(ules) 
? Thanks for hints. :-)

I have CSV file and I have trouble using parse one-liner. The case 
is, that I export tel. list from Lotus Notes, then I save it in Excel 
into .csv for rebol to run thru. I wanted to use:

foreach line ln-tel-list [append result parse/all line ";"]


... and I expected all lines having 7 elements. However - once last 
column is missing, that row is incorrect, as rebol parse will not 
add empty "" at the end. That is imo a bug ...

I used csv = semicolon separated values and no quotes in-there ...

yes, it is, as I expect all lines having 7 elements ... once there 
is not such an element, I can't loop thru result ... well, one condition 
will probably solve it, but imo it is a gug .... rebol identifies 
;; and puts "" inthere, but csv, at the end, will use "value;", and 
rebol does not count that ...

append/only will not help, result of parse will varry, and it should 
not ...

append/only will, because pick returns none if a column is not present, 
and set works with that too

but I like to use flat structure and foreach [real name of vars here], 
so I need consistent record length :-)

oh no, I am at the ends ... so bye bye beautifull oneliners ... I 
just found item which contains set of quotes :-) rebol will not translate 
that and my block is confused once again :-)

damned, I will delete that contact and I will be done ;-) One contact 
from 3 000 does not count, even if that is secreatery of general 
director :-)

How many possibilities are there for the values "team one" and "team 
two"?  Could you use a team repository? --

Ingo I made simple test and it seems it's something like 512

I have some problems to translate RULE based on PERL  regular expression 
to rebol parse rules. Maybe this group can help me. For example:
PERL REGULAR EXPRESSION RULE:
[[ c ]] (e|è|é|ê|i|î|y) -> s 

 we need to match the #"c" char and test  the next char to know if 
 we trap S phonem. example "Ce" must be emited S et E. I plan to have 
 2 rules rule: [ .../... "c" etc...  | "e"  etc... ]

More complicated rule:
an [[ c ]] T ->	
like for "banc"

you know p: and :p in parse-rules?

and your code is mixing rebol code with the parse dialect.

oops .. should be "2 skip" and not "skip 2"

That *is* a nice and easy solution, thanks a lot!

Working on a textile parser over here to build my 'parse skills and 
make it easier to build my website with Rebol

and all of my pages are already in textile format, and I think a 
few of my friends would be more interested in Rebol if I had a textile 
parser for them

Volker thank you it works great now and the code rule is tiny ;)

you probably need to count them and see where you end up after finding 
all parens. I'm not sure if it can be used to see which are missing...

count one up on #"(" and one down on #")". If correct, the end result 
is zero.

You can make it a little more complicated to add more markup types, 
but the basic structure is the same. The trick is the :a before the 
paren - otherwise it won't work, and you can crash older versions 
of REBOL.

That works nicely too! I'll look more at NicomDoc later, but BrianH's 
tip makes tests for "*test*" and "*test" pass

No nesting, but with a little recursion and different start and end 
tags, this can be adapted to handle that too.

If you want to determine whether there have been any replacements, 
change the second any to some and parse will return true only when 
replacements have been made. Be careful to avois use of the markup 
characters in your replacement text.

Tomc, that will crash older versions of REBOL, and not work on newer 
versions. You need to reset the parse position to before the change, 
before the paren where you make the change. Otherwise parse will 
be referencing a point off the end of the string at the end of the 
paren, before you can reset it. This used to crash REBOL so bad the 
interpreter disappeared.

yes, shortening the string you are parsing would pull the rug out 
from under the interperter,
(and I was aware that the string was being lengthened) 

note: setting the parse pointer back to :here will position you before 
the "*"  

you may be better off  with :here skip to gaurentee progress in the 
case the change fails

OK, I tried this: parse "abc" [to "bc" a: "bc" (change/part a "b" 
2)]

It returns true on View 1.3 and Core 2.6, but false on View 1.2 and 
Core 2.5.0.

Here's a simplified version of my example that can handle multiple 
instances of multiple markup types and be adapted to different end 
tags (thanks Tomc for the idea!):

markup-chars: charset "*~"
non-markup: complement markup-chars
tag1: ["*" "<strong>" "~" "<i>"]
tag2: ["*" "</strong>" "~" "</i>"]
parse/all data [
    any non-markup
    any [

        ; This next block can be generated if you have many markup types...

        [a: copy b "*" copy c to "*" copy d "*" e: | a: copy b "~" copy c 
        to "~" copy d "~" e: ]
        :a (change/part a rejoin [tag1/:b c tag2/:d] e)
        any non-markup
    ]
    to end
]

The intent of the licence is to prevent on-selling a REBOL-like took 
which does little more than expose the REBOL API. This is no different 
from other [commercial] programming languages where they don't want 
you buying a licence for $2,000 and then distributing essentially 
the same functionality (i.e. a language / compiler) for $5.


An easy way to express this notion is: "You are not permitted to 
use this product to create a product that could be construed by a 
reasonable person to be the same or a derivative." Finding the reasonable 
person is the trick! ;)

so even something as simple as
<%
	print {hello
%>

kills rsp and subsequent rsp

I believe the icons are needed because I was experimenting with it 
and found without it I couldn't not get the results I wanted on XP.

ie. this fails ... 

read what-dir

and you have to 

read dirize what-dir

instead?

IconSushi.  It's not so much an editor as a compiler, importing PNG 
(including 32-bit) and BMP...

Hmm, I tracked down pngcrush (http://sourceforge.net/project/showfiles.php?group_id=1689&package_id=6641), 
 and while its usage seems clear enough:

	pngcrush.exe -bit_depth 4 24bit.png 4bit.png

it dosen't seem to work.

Not quite the batch solution I had in mind, but MSPaint lets you 
save a PNG as a 4, 8 or 24 bit BMP ... which is good enough for IconSushi 
... which works with good old ResHacker ... which lets me, wait for 
it, change the default REBOL icons in an encapped program ... and 
folks wonder why REBOL isn't mainstream yet! ;) (An "Encap: [icons 
%my-icons.ico]" type option would be a good thing IMHO).

Has anyone seen his message before?  I tried typing into a view application 
under windows 2000, and got this


The ordinal 6467 could not be located in the dynamic link library 
MFC42.DLL

240 kb kernel... is this completely barebones and is this bigger 
or smaller than R2?

The idea is to open source a lot, and just keep the low level stuff 
closed, but RT will post something official.

so now Ruby can finally have their View clone? :-) They will take 
rebol.dll, create one function "do-rebol", and voila :-)

Gabriele - this is actually good, you know? First Ruby guys will 
think they got View for Ruby, next they will realise, they don't 
actually need Ruby, to just start and use Rebol, so they will drop 
Ruby :-)

Henrik -- Maxim has mentioned your styles for Glayout several times, 
and demoed them.

Said he was up at 04:00 last night, and it took him 20minutes to 
integrate your styles into Glayout.

along with coders like Anton, Volker and Henrik ...

nah, we'll do our own thing and don't want external influences ;)

me for sure and I hope someone teach me surfing.. 8-)

Carl's comments on merging of VID dialect and global namespace. It 
took me a long while to learn it, but I appreciate it more and more.

Especially as lunch includes free wine and coffee :-)

Free wine?!?  Don't let Carl near it, he will start comparing it 
to his own Vineyards...and it will be a 3-4 hour lunch!

and they'll all be drunk, which won't look good on the live stream

obviously there must be a solution to this problem already, but I'm 
just thinking two mics, one soft, one loud, and a "compressor" box 
that selects the loudest signal which does not clip.

Also cuts down on pink noise and static.

It is pretty good given teh room (which is all walls and windows, 
not much to absorb sound)

So the mic is hot, and the room is hot TOO!

Yes, putting foam over the mic would help a little, but it would 
also mean that it would not work as well iat a distance.  What you 
really want is a mic wired on the person (which has foam built in), 
 and no mic on the room.

So you localize the sound and the source.

I bought a really cheap mic recently - I think it was $3 - and the 
frequency range is abysmal.

Mics have dropped in price, just like optics for cameras...but $3 
is still TOO CHEAP.  Pay $50+ and you will get a good mic.  I always 
trust Plantronics for headsets.

I have bougfht maybe several hundred mics and headsets over teh yers.

The ER 4 are pretty good.  And I like traveling with them.

geomol, I just put the address into the URL in quicktime and it worked.

Watching this on Safari/Quicktime I get sound and a green image?

Best I can tell, if yoru CPU gets bogged with other stuff (opening 
other windwos, virus checkers, etc.) then it loses sync, and you 
have to refresh.

OK, I'm done looking at Richard and Gab, move 45' left. : )

And that he has a name badge!!!!

Anton - me too - I sent google to Rebol.com and net and guessed, 
but no luck :-)

ge ahead and try
 about the server

I reloaded and it worked.

(btw docbase is already running, as carl said this morning, and carl 
has been writing a lot of things in it.)

considers allowing object! to define the port handler... like a feel.. 
.MUCH easier to Tweak and scan handler