recipe/rc2014/selfhost: new recipe

It's now a simple flag updated by the C< implementation. No more
routine pointer.

The goal is to avoid tons of "ok" during a LOAD operation.

blk/083

@@ -1,5 +1,6 @@
-C<?* is a pointer to a word being called by C<?. If 0 or 1,
-will return that value as-is.
+C<? is a flag indicating whether a character is waiting in the
+input stream. 1 means yes, 0 means no. It is the responsibility
+of C<* to update that flag.
 
 WORDBUF is the buffer used by WORD
 
@@ -12,5 +13,4 @@ jump to this address. If you use one of those slots for an
 interrupt, write a jump to the appropriate offset in that RAM
 location.
 
-
                                                         (cont.)

blk/262

@@ -1 +1 @@
-263 LOAD 265 LOAD
+263 LOAD 265 LOAD 268 LOAD

blk/267

@@ -0,0 +1,6 @@
+XPACK - pack source code
+
+The goal of this word is to pack source code in tight places,
+such as on the boot section of an EEPROM. It takes a block
+number, reads it and packs it to HERE. It normalizes all
+whitespaces to a single space and ignore comments.

blk/268

@@ -0,0 +1,16 @@
+: XPACK ( blkno -- )
+    BLK@
+    BLK( 0x2e RAM+ ! ( boot ptr )
+    ['] (boot<) 0x08 RAM+ ! ( C<* override )
+    BEGIN
+        WORD
+        0x2e RAM+ @ BLK( 1024 + < IF
+            DUP LIT< ( S= IF
+                DROP [COMPILE] (
+            ELSE
+                SCPY 0x20 H@ 1- C!
+            THEN 0 ( loop again )
+        ELSE 1 ( stop looping ) THEN
+    UNTIL
+    0 0x08 RAM+ !
+;

blk/398

@@ -2,7 +2,7 @@
   for an abort message )
 : (parse) (parsed) NOT IF ABORT THEN ;
 
-: C<? 0x06 RAM+ @ DUP 2 > IF EXECUTE THEN ( 06 == C<?* ) ;
+: C<? 0x06 RAM+ @ ;
 : C<
     0x08 RAM+ @  ( 08 == C<* override )
     DUP NOT IF DROP 0x0c RAM+ @ THEN  ( 0c == C<* )

forth/blk.fs

@@ -61,17 +61,21 @@
 ;
 
 : LOAD
-    ( save BLK>, C<* override and boot< ptr to RSP )
+    ( save restorable variables to RSP )
     BLK> @ >R
     0x08 RAM+ @ >R
-    0x2e RAM+ @ >R
+    0x06 RAM+ @ >R  ( C<? )
+    0x2e RAM+ @ >R  ( boot ptr )
     BLK@
     ( Point to beginning of BLK )
     BLK( 0x2e RAM+ !
     ( 08 == C<* override )
     ['] _ 0x08 RAM+ !
+    ( While we interpret, don't print "ok" after every word )
+    1 0x06 RAM+ !  ( 06 == C<? )
     INTERPRET
     R> 0x2e RAM+ !
+    R> 0x06 RAM+ !
     ( Before we restore C<* are we restoring it to "_"?
       if yes, it means we're in a nested LOAD which means we
       should also load back the saved BLK>. Otherwise, we can

forth/core.fs

@@ -18,7 +18,8 @@
 : BEGIN H@ ; IMMEDIATE
 : AGAIN COMPILE (br) H@ - , ; IMMEDIATE
 : UNTIL COMPILE (?br) H@ - , ; IMMEDIATE
-: ( BEGIN LIT< ) WORD S= UNTIL ; IMMEDIATE
+: _ BEGIN LIT< ) WORD S= UNTIL ; IMMEDIATE
+40 CURRENT @ 4 - C!
 ( Hello, hello, krkrkrkr... do you hear me?
   Ah, voice at last! Some lines above need comments
   BTW: Forth lines limited to 64 cols because of default
@@ -27,6 +28,8 @@
   "_": words starting with "_" are meant to be "private",
   that is, only used by their immediate surrondings.
 
+  40 is ASCII for '('. We do this to simplify XPACK's task of
+  not mistakenly consider '(' definition as a comment.
   LITS: 34 == litWord
   LITA: 36 == addrWord
   COMPILE: Tough one. Get addr of caller word (example above

forth/parse.fs

@@ -54,7 +54,7 @@
     DUP @ 25136 = NOT IF 0 EXIT THEN    ( a 0 )
     ( We have "0b" prefix )
     2+
-    0 ( a r )
+    0  ( a r )
     BEGIN
     SWAP C@+                               ( r a+1 c )
     DUP NOT IF 2DROP 1 EXIT THEN           ( r 1 )

forth/readln.fs

@@ -59,15 +59,18 @@
     LF IN( IN> !
 ;
 
-: RDLN<? IN> @ C@ ;
-
-( And finally, implement a replacement for the (c<) routine )
+( And finally, implement C<* )
 : RDLN<
-    RDLN<?                      ( c )
-    ( not EOL? good, inc and return )
-    DUP IF 1 IN> +! EXIT THEN   ( c )
-    ( EOL ? readline. we still return typed char though )
-    (rdln)                      ( c )
+    IN> @ C@
+    DUP IF
+        ( not EOL? good, inc and return )
+        1 IN> +!
+    ELSE
+        ( EOL ? readline. we still return null though )
+        (rdln)
+    THEN
+    ( update C<? flag )
+    IN> @ C@ 0 > 0x06 RAM+ !  ( 06 == C<? )
 ;
 
 ( Initializes the readln subsystem )
@@ -78,7 +81,7 @@
       the last typed 0x0a and one for the following NULL. )
     INBUFSZ 4 + ALLOT
     (infl)
-    ['] RDLN<? 0x06 RAM+ !
     ['] RDLN< 0x0c RAM+ !
+    1 0x06 RAM+ !  ( 06 == C<? )
 ;
 

recipes/rc2014/README.md

@@ -25,7 +25,7 @@ are other recipes related to the RC2014:
 
 * [Writing to a AT28 from Collapse OS](eeprom/README.md)
 * [Accessing a MicroSD card](sdcard/README.md)
-* [Assembling binaries](zasm/README.md)
+* [Self-hosting](selfhost/README.md)
 * [Interfacing a PS/2 keyboard](ps2/README.md)
 
 ## Recipe

recipes/rc2014/run.fs

@@ -1,7 +1,2 @@
-: (c<) KEY DUP EMIT ;
-: _
-    ACIA$
-    ." Collapse OS" CRLF
-    ( 0c == CINPTR )
-    ['] (c<) 0x0c RAM+ !
-; _
+: x KEY DUP EMIT ;
+: _ ACIA$ (ok) ['] x 0x0c RAM+ ! ; _

recipes/rc2014/sdcard/README.md

@@ -105,5 +105,22 @@ And thats it! You have full access to disk block mechanism:
 
 (at this moment, the driver is a bit slow though...)
 
+## How do I fill my SD card with Collapse OS' FS?
+
+Very easy. You see that `/emul/blkfs` file? You dump it to your raw device.
+For example, if the device you get when you insert your SD card is `/dev/sdb`,
+then you type `cat emul/blkfs | sudo tee /dev/sdb > /dev/null`.
+
+## What to do on SDerr?
+
+If you get `SDerr` in the middle of a LOAD operation, something went wrong with
+the SD card. The bad news is that it left your xcomp operation in an
+inconsistent state. If your at the beginning of it, it's easier to restart it
+entirely.
+
+If you're towards the end, you might want to repair it. To do so, you'll have to
+bring your `XCURRENT` and `HERE` values to where they were before the LOAD
+operation. You could have thought ahead and printed them before the LOAD, but if
+you didn't, you'll just have to dig in your memory with `DUMP`.
 [schematic]: spirelay/spirelay.pdf
 [inspiration]: https://www.ecstaticlyrics.com/electronics/SPI/fast_z80_interface.html

recipes/rc2014/selfhost/README.md

@@ -0,0 +1,54 @@
+# Assembling Collapse OS from within it
+
+This is where we tie lose ends, complete the circle, loop the loop: we assemble
+a new Collapse OS *entirely* from within Collapse OS and write it to EEPROM,
+either for another RC2014 or for an OS upgrade.
+
+## Gathering parts
+
+* stage4 from `sdcard` recipe. If you want to write to EEPROM as the final step,
+  you'll need a hybrid stage4 that also includes stuff from the `eeprom` recipe.
+
+## Building stage 1
+
+### Part 1
+
+Building the first part of stage 1 (the binary part, before the inlined-source
+part) from within Collapse OS is actually very similar from building it from a
+modern environment. If you take the time to look at the base recipe `Makefile`,
+you'll see `cat xcomp.fs | $(STAGE2)`. That command builds part 1. Open
+`xcomp.fs` in a text editor and take a look at it.
+
+To assemble stage 1 from RC2014, all you need to do is to type those commands
+in the same order, and replace the `H@ 256 /MOD 2 PC! 2 PC!` lines with `H@ .X`.
+Those commands will inform you of the begin/end offsets of the assembled binary.
+
+The meaning of these commands is not explained here. You are encouraged to read
+the in-system documentation for more information.
+
+However, one thing you should know is that because the SD card driver is a bit
+slow, some of these commands take a long time. Multiple minutes. Be patient.
+
+Once all your commands are run and that you have your begin/end offset (write
+them down somewhere), you're ready to assemble part 2.
+
+### What to do on SDerr?
+
+If you get `SDerr` in the middle of a LOAD operation, something went wrong with
+the SD card. The bad news is that it left your xcomp operation in an
+inconsistent state. If your at the beginning of it, it's easier to restart it
+entirely.
+
+If you're towards the end, you might want to repair it. To do so, you'll have to
+bring your `XCURRENT` and `HERE` values to where they were before the LOAD
+operation. You could have thought ahead and printed them before the LOAD, but if
+you didn't, you'll just have to dig in your memory with `DUMP`.
+
+You're looking at the offset of the last wordref of the *previous* LOAD
+operation. That offset is going in `XCURRENT`. Then, you're looking at the end
+of that word. That offset goes in `HERE`. Once you've done that, relaunch your
+LOAD.
+
+### Part 2
+
+TODO

tools/stripfc.c

@@ -13,7 +13,7 @@ by more than once space or by a newline. Hackish, but works.
 
 int main()
 {
-    int spccnt = 2; // if the first char is a (, consider it a comment opener.
+    int spccnt = 1; // if the first char is a (, consider it a comment opener.
     int incomment = 0;
     int c;
     c = getchar();
@@ -24,7 +24,7 @@ int main()
                 // doesn't like when they're not there...
                 putchar(c);
             }
-            spccnt += 2;
+            spccnt += 1;
         } else if (c == ' ') {
             spccnt++;
         } else {
@@ -33,7 +33,7 @@ int main()
                     incomment = 0;
                 }
             } else {
-                if ((c == '(') && (spccnt > 1)) {
+                if ((c == '(') && spccnt) {
                     putchar(' ');
                     spccnt = 0;
                     int next = getchar();