Detect overwriting a non-zero value; optionally suppress the check.

  * src/aes.adb: add -z/--overwrite-nonzero option, save to Store.

  * src/analytical_engine-card.ads (Image): new.
  * src/analytical_engine-card.adb: likewise.
    (Zero): deleted.
    (Execute(Variable_Card)): use new version of Store.Get.
    (Image): new.

  * src/analytical_engine-card_reader.adb (Execute): name the loop
      Execution. Use Card.Image instead of d.i.y. Catch and report
      exceptions.

  * src/analytical_engine-store.ads (Allow_Overwrite_Nonzero): new.
    (Get): new parameter Preserve.
    (Store_Error): new.
  * src/analytical_engine-store.adb (Zero): new.
    (Overwrite_Nonzero_Allowed): new.
    (Allow_Overwrite_Nonzero): new.
    (Set): if required, check for overwriting non-zero column, raise
      Store_Error if so.
    (Get): handle new Preserve parameter; if tracing, report if column
      has been zeroed; if not set, set the column to zero.

  * bernouilli.ae: load-and-zero where required (Z instead of L).
  * bernouilli5.ae: likewise.

  * README.md: updated.

Author: simonjwright

README.md

@@ -32,7 +32,8 @@ The Ada emulator is written using [Ada2012][], and uses the
 [GMP binding in GNATCOLL][].
 
 If you have the [FSF GCC 6.1.0 binary for OS X][] a suitable library
-is provided. GMP and GNATCOLL are reasonably straightforward to build.
+is provided. Otherwise, GMP and GNATCOLL are reasonably
+straightforward to build.
 
 [Ada2012]: http://www.ada-auth.org/standards/rm12_w_tc1/html/RM-TOC.html
 
@@ -42,6 +43,32 @@ is provided. GMP and GNATCOLL are reasonably straightforward to build.
 
 [FSF GCC 6.1.0 binary for OS X]: https://sourceforge.net/projects/gnuada/files/GNAT_GCC%20Mac%20OS%20X/6.1.0/
 
+## Running
+
+There are three switches:
+
+  * `-h` or `--help`: outputs help.
+  * `-t` or `--trace`: trace execution (like card T1)
+  * `-z` or `--overwrite-nonzero`: allow storing to a non-zero column
+
+and, as usual, run by
+
+    ./aes [switches] [card-chain-file]
+
+(uses standard input if no card chain file is supplied).
+
+The "storing to a non-zero column" issue is referenced in [this paper][]
+by Rainer Glaschick, section 3.5, Memory peculiarities. It would have
+been necessary to ensure that a column that was to be written to was
+zero beforehand; the operation of storing a value involved racks
+transferring the value on the source to the destination as the source
+digits were rotated back to zeros.
+
+If `-z` is given, overwriting will be allowed; otherwise, an error
+will be reported ane exceution will halt.
+
+[this paper]: http://rclab.de/rclab/_media/analyticalengine/aal_noteg_glaschick_v1.2.pdf
+
 ## Programming Cards
 
 The design here is as in the [Fourmilab Java implementation][]. To
@@ -118,12 +145,13 @@ The [Deep Learning with the Analytical Engine][] project, with the
 > sufficiently long).
 
 The program consists of over 400,000 cards! (it was generated by
-Python scripts from a high-level description). As a timing test, the
-program was run for the first 10 images only: in Java, this took 18
-seconds, of which 10 seconds was initial processing, in Ada (where the
-initial processing was about a second) with `-O0 -gnata` 44 seconds,
-with `-O2 -gnatp` 22 seconds. So the Ada implementation is about 3
-times slower than the Java one. More work to be done!
+Python scripts from a high-level description). It requires `-z` to
+run. As a timing test, the program was run for the first 10 images
+only: in Java, this took 18 seconds, of which 10 seconds was initial
+processing, in Ada (where the initial processing was about a second)
+with `-O0 -gnata` 44 seconds, with `-O2 -gnatp` 22 seconds. So the Ada
+implementation is about 3 times slower than the Java one. More work to
+be done!
 
 [Deep Learning with the Analytical Engine]: https://cp4space.wordpress.com/2016/02/06/deep-learning-with-the-analytical-engine/
 [Deep Learning with the Analytical Engine repository]: https://gitlab.com/apgoucher/DLAE

bernouilli.ae

@@ -10,7 +10,7 @@
  A bug in operation 4 has been fixed (the comment requires V4/V5,
  AAL's code does V5/V4).
 
- Another bug is at operation 24: the value accumulated in V13 is *the
+ Another bug was at operation 24: the value accumulated in V13 is *the
  negative* of the required value, see Note G Equation 9.
 
  useful constants
@@ -38,13 +38,13 @@ s006
 
  operation 2
 -
-l004
+z004
 l001
 s004
 
  operation 3
 +
-l005
+z005
 l001
 s005
 
@@ -57,13 +57,13 @@ s011'
 
  operation 5
 /
-l011
+z011
 l002
 s011'
 
  operation 6
 -
-l013
+z013
 z011
 s013
 
@@ -85,7 +85,7 @@ l006
 l007
 s011'
 
- operation 10
+ operation 10, using b1 (v021)
 *
 l021
 l011
@@ -95,25 +95,25 @@ s012
  operation 11
 +
 z012
-l013
+z013
 s013
 
  operation 12
 -
-l010
+z010
 l001
 s010
  --------------------
  operation 13
 -
-l006
+z006
 l001
 s006
 
  operation 14
 +
 l001
-l007
+z007
 s007
 
  operation 15
@@ -126,20 +126,20 @@ s008'
  operation 16
 *
 z008
-l011
+z011
 >10
 s011
 
  operation 17
 -
-l006
+z006
 l001
 s006
 
  operation 18
 +
 l001
-l007
+z007
 s007
 
  operation 19
@@ -152,11 +152,11 @@ s009'
  operation 20
 *
 z009
-l011
+z011
 >10
 s011
 
- operation 21
+ operation 21, using b3 (v022)
 *
 l022
 l011
@@ -166,25 +166,25 @@ s012
  operation 22
 +
 z012
-l013
+z013
 s013
 
  operation 23
 -
-l010
+z010
 l001
 s010
  .................... repeat starts here
  operation 13
 -
-l006
+z006
 l001
 s006
 
  operation 14
 +
 l001
-l007
+z007
 s007
 
  operation 15
@@ -197,20 +197,20 @@ s008'
  operation 16
 *
 z008
-l011
+z011
 >10
 s011
 
  operation 17
 -
-l006
+z006
 l001
 s006
 
  operation 18
 +
 l001
-l007
+z007
 s007
 
  operation 19
@@ -223,11 +223,11 @@ s009'
  operation 20
 *
 z009
-l011
+z011
 >10
 s011
 
- operation 21 (NB using v023 this time)
+ operation 21 (NB using b5, v023 this time)
 *
 l023
 l011
@@ -237,12 +237,12 @@ s012
  operation 22
 +
 z012
-l013
+z013
 s013
 
  operation 23
 -
-l010
+z010
 l001
 s010
  --------------------
@@ -257,8 +257,8 @@ p
  operation 25
 +
 l001
-l003
+z003
 s003
- -- and clear V6, V7
-n006 0
-n007 0
+ -- and clear V6, V7 (should have been done earlier, operation 19)
+ n006 0
+ n007 0