Dar Users,

1. The next version correct the behavior of releasing local symbols. Module now keeps the generation number of a symbol at its exist. It used to reset the symbols to generic, but for instance, for a usage like as:

Module[{x},
f[x_] := x];

the local symbol x$n (n: generation) should be reserved after the exit of Module. If this change cause any side effect, please let me know.

Dear Users,

1. The memory allocation in EXPAND in main will use the same function as FFS in the next version to hopefully reduce a chance of accidental allocation in an inaccessible space.




> Dar Users,
>
> 1. The next version correct the behavior of releasing local symbols. Module now keeps the generation number of a symbol at its exist. It used to reset the symbols to generic, but for instance, for a usage like as:
>
> Module[{x},
> f[x_] := x];
>
> the local symbol x$n (n: generation) should be reserved after the exit of Module. If this change cause any side effect, please let me know.

> Dear Users,
>
> 1. The memory allocation in EXPAND in main will use the same function as FFS in the next version to hopefully reduce a chance of accidental allocation in an inaccessible space.
>
かなり特殊なケースですが、構文上は合法かつエラー/警告は出てない入力での
挙動がデグレードしています。(前のVersionの挙動もバグ臭いですが…)

* 既存のバグくさい挙動
V1.0.10.4.16a20だと
% /SAD/bin/gs /ldata/SuperKEKB/Lattice/HER/herfqlc5604.sad
FFS USE=ASCE;
VISIT ASCE
VISIT ASCE
BYE
BYE
でtfreeがメッセージを出した後、スタックします(FFSプロンプトに戻らない)
一回目のBYEで、 "tfree: detect null pointer"を一回出力
二回目のBYEで、 "tfree 4643520 1"を大量に出力

V1.0.10.4.17a11だと
% ./bin/gs /ldata/SuperKEKB/Lattice/HER/herfqlc5604.sad
FFS USE=ASCE;
VISIT ASCE
VISIT ASCE
BYE
BYE
で、Segmentation Faultが発生
スタックが、Segmentation Faultに変化
コアが取得出来る分、デバックはしやすくなった？

* デグレード
V1.0.10.4.17a11にて
% ./bin/gs /ldata/SuperKEKB/Lattice/HER/herfqlc5604.sad
FFS USE=ASCE;
VISIT ASCE
VISIT ASCE
BYE
DISP
や
% ./bin/gs /ldata/SuperKEKB/Lattice/HER/herfqlc5604.sad
FFS USE=ASCE;
VISIT ASCE
BYE
DISP
が Segmentation Fault

V1.0.10.4.16a20だと Segmentation Faultにならないのでデグレードと思われるが、
元々 VISIT*2/BYE*2は動かないようなので、たまたま動いているだけか？

Dear Users,

1. A bug was found in definition of Shared in init.n to wrongly define the global symbol v .

Dear Users,

1. The generation of symbol for a pattern was probably inappropriately handled. It is corrected, but any side effect is possible in existing scripts.

> Dear Users,
>
> 1. The generation of symbol for a pattern was probably inappropriately handled. It is corrected, but any side effect is possible in existing scripts.

SAD Ver.1.0.10.4.17a20でscript/design_example.sadを走らせると、描画に失敗します。
SAD Ver.1.0.10.4.17a11では、ちゃんと描画できています。
　共にMacOSLion上でコンパイルおよび実行をしました。

試しに、
ListPlot[Range[10]]; Update[]
をTk$Echo = 1にて Tcl/Tk側への出力を差分してみましたが、

本質的に、異なっている部分は、
* Frame Tick描画時の 座標及び 文字列
だけでした。

ListPlot[]が返すGraphicsオブジェクト自体のPrint[]結果は同一なので、
Canvas.n内部が怪しいように思われます

f[x_, x_] などのように、同一パターンが含まれる場合のmatchingに失敗しています。調査中です。

Dear Users,

1. The bug introduced by 1.0.10.4.17a20 to wrongly handle equivalent patters are corrected. Still watch more side effect!

Dear Users,

1. An issue was found in setting a real value to a local symbol of Module (tvset). It may have caused some unpredictable behavior since the change in 1.0.10.4.17a. The original problem in tvset was introduced probably in 1999, but did not reveal the symptom until 1.0.4.17a.