by

Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Activity
  • Oct 20 2019 22:59
    @dockimbel banned @SmackMacDougal
  • Dec 03 2017 05:53
    @PeterWAWood banned @matrixbot
  • Sep 28 2016 12:19
    @PeterWAWood banned @TimeSeriesLord
  • Aug 13 2016 03:23
    @PeterWAWood banned @Vexercizer
Vladimir Vasilyev
@9214
@GiuseppeChillemi click on timestamp of the message you're interested in and copy its link from the address bar. Like this:
https://gitter.im/red/help?at=5acd1de3080a3850531a9b18
GiuseppeChillemi
@GiuseppeChillemi
@9214 I have clicked everywhere but timestamp ! Thanks !
GiuseppeChillemi
@GiuseppeChillemi

@9214

unnamed0.png

@nedzadarek what you seem to call a "variable" is a binding + relation between symbol ID and value inside context.

What is SYMBOL ID ? I am getting confused. Could you make a graph adding real words, blocks and arrows to this structure and also underliyng C like rappresentation of the data structure ?

Vladimir Vasilyev
@9214

Word reference its context through a node (just a special intermediate structure) and has a symbol ID, which is an index in a symbol table. Context, in turn, is just a two-column table: first column is for symbol! slots, second one is for any-value! slots.

When a word is evaluated, a value is fetched in the context pointed by that word, from the value slot that correspond to the symbol ID held by the word. In that sense, words is an instance of an internal symbol! datatype: word = a context reference + a symbol ID.

Symbol table, as I understand it, is nothing more than block of symbol! cells, one per Red session currently (should be one per module in the future). Symbol ID is an index of a cell in that block. symbol! cell holds a UTF-8 representation of a word (its "spelling", it's actually a pointer to a string buffer, again through node) and an alias field. If word is an alias of the other word (think case-insensitivity), this field will hold an index of the original word.

Ryan Cole
@prapro
Basically it's The Force. https://youtu.be/x2YQJsbbWNA
Vladimir Vasilyev
@9214
https://gitter.im/red/help?at=5acd2aa75f188ccc154fc0b2
:point_up: I think it should be " ... two-column table: first column is for symbol IDs ..." instead, but the idea is the same.
BeardPower
@BeardPower

The data of a type struct is stored in a cell, which is a red-value. These structures are 128bit slices on the stack (4 32bit integers). A cell consists of
header
data1
data2
data3

Which are filled/type-casted to the appropriate types.

Vladimir Vasilyev
@9214
Cells and slots are synonyms, though "cell" is used as a low-level term. Cells are allocated from the heap as block! buffers.
BeardPower
@BeardPower
It could easily confuse developers, as the compiler is printing error messages in regards to cells (data1, data2 data3) instead of the real struct, which it's casted to.
If e.g. a type is using data2 for it's value, the error message would print "error in data2" instead of "error in value".
Well, isn't a Cell consisting of 4 32bit slots? The stack consists of cells, which are build from slots.
Vladimir Vasilyev
@9214
@BeardPower but you can look at the header with type ID in it, right?
BeardPower
@BeardPower
@9214 Yes, but you still have to look up the fields by hand.
The error message will not print out the ID.
Vladimir Vasilyev
@9214

@BeardPower per my notes

Both Red and Rebol use arrays of 128-bit cells (aka slots) to store values (32-bit header + 96-bit payload).

BeardPower
@BeardPower
The error handling could be better. If an action is called internally (e.g. form and mold to print out the result of a type-cast in the REPL), but is null, it will show a stack-trace with failed form and mold calls. It would be easier, if there is a error message like: form/mold needs to be implemented!
Vladimir Vasilyev
@9214
Though maybe you're right, initially I thought that cell = 4 slots too.
BeardPower
@BeardPower

@9214 Yes, you are right. A cell is a red-value, which is 128-bit, but the allocator has this comment: placeholders to make a 128-bit cell

So I asked myself, if there are lower bit cells, and the slots are 128bit cells.

cell!: alias struct! [
    header    [integer!]                        ;-- cell's header flags
    data1    [integer!]                        ;-- placeholders to make a 128-bit cell
    data2    [integer!]
    data3    [integer!]
]
Vladimir Vasilyev
@9214
@BeardPower 1 header (32 bits) and 3 placeholders (96 bit payload) = 1 cell (128 bits total).
BeardPower
@BeardPower
So the "cell" moved onto the stack is then called a slot :)
Vladimir Vasilyev
@9214
:mortar_board:
BeardPower
@BeardPower
The cell is in the stack's slot.
Vladimir Vasilyev
@9214
Very slot, should allocate.
BeardPower
@BeardPower
:smile:
@9214 Are your notes online? Doing double work is :older_man:
Trying to siphon every info to prepare for Red 2.0 ;-)
Vladimir Vasilyev
@9214
@BeardPower don't siphon - mediate.
Inhale, exhale, do the "ohhhmmmm" sound, stare at the sources and slowly whirl your pinky finger around the belly button.
Counter-clockwise!
BeardPower
@BeardPower
Feeling :dizzy_face: Seems to work ;-)
redredwine
@redredwine

@9214 Thanks for the parse rule. it works. I was trying to use OPT, but could not figure how to use OPT.

collect [
some [
keep["tid=" to ". "]
keep[to "startTime:" ]
keep[ to ", "]
opt [to "endTime:"]
opt[ keep [to ", "]]
[thru newline]
]

the above is what i was trying to get working, as soon as I added the OPT lines, it messed up everything.
nedzadarek
@nedzadarek

@lepinekong_twitter wait... print doesn't print?!
@9214 I disagree about some stuffs but I don't want to argue. Thank you for your time.

stick to our conventions and wordings

Pun intended, right?

Alexander Baggett
@AlexanderBaggett
Hey guys, I am trying to get some code that doc gave me to work. Latest automated build doesn't take it. Am I doing something wrong, or has Rich text not been merged into master yet?
view [rich-text data [i b "Hello" /b font 24 red " Red " /font blue "World!" /i]]
Alexander Baggett
@AlexanderBaggett
nm it looks like the rich-text branch is still active and not yet merged into master
ne1uno
@ne1uno
AlexanderBaggett it's still in the rich-text branch, it's not hard to build the console
Alexander Baggett
@AlexanderBaggett
Honestly I have never tried to build the console. What's involved?
ne1uno
@ne1uno
grab rebol2 view from rebol.org. I think the instructions are in the readme displayed on the github page
you can get the source of the branch from github as a zip or use some git client
Vladimir Vasilyev
@9214
@redredwine opt either matches the rules or not, but advances the parsing in any case. However, in your example missing endTime won't be substituted by none and you'll stuck with variable-length of tokens per line.
Vladimir Vasilyev
@9214

@nedzadarek so, now you disagree with technical details and objective facts? Whatever.
https://www.youtube.com/watch?v=b47wP_yMCf4

Words and variables are different both in their implementations and at conceptual level. Words can serve as variables in a very, very special sense (to the degree that you can't just call them "variables" without wasting your breath on 1-2 hour explanation of difference between the two) , but they are not variables, neither technically (they differ in implementations) nor conceptually (we use "words" and "bindings" for a reason!).

Vladimir Vasilyev
@9214
@redredwine or maybe your rule will keep something that is not a timestamp, can't remember the details of log format now.
hiiamboris
@hiiamboris
@9214 I've found your explanation of words vs variables enlightening.. esp. that graph :+1:
Vladimir Vasilyev
@9214
@hiiamboris you're welcome! I hope I didn't spread any BS or misinformation here. :)
hiiamboris
@hiiamboris
@9214 can a word have absolutely no context? not even refer to system/words?
as an illustration:
>> system/words = context? first [pewpew]
== true
Boleslav Březovský
@rebolek
it can in Rebol3, not sure about Red
Vladimir Vasilyev
@9214
@hiiamboris it cannot in Red, at least for now, but that might change with modules.