Can we invoke a subroutine from a subroutine? Absolutely! In fact, none of the mechanism discussed so far has to change.
The trace of this algorithm is as follows (the first column identifies a row in the trace for easier reference):
trace | A | B | C | D | explanation |
| 1 | line# | x |
|
||
| 2 | pre | ? |
|
||
| 3 | 8 | 0 |
| ||
| 4 | 9 | retline# | At this point, column C is the leftmost available column. |
||
| 5 | 10 |
| |||
| 6 | 5 | retline# | At this point, column C is already in use. As a result, when “s1” is invoked from “s2”, column D is the leftmost available column. |
||
| 7 | 6 |
| |||
| 8 | 2 | 1 | |||
| 9 | 3 | retline# | At the end of a subroutine, the rightmost allocated “retline#” is used to indicate where to continue execution. In this case, we continue execution on line 6. |
||
| 10 | 6 | 11 |
| ||
| 11 | 7 | retline# | Because column D is already deallocated, column C becomes the rightmost “retline#” column. As a result, we continue execution on line 10 in the next row of the trace after column C is deallocated. |
||
| 12 | 10 | retline# | At this point, column C is already deallocated (by the previous row of the trace). As a result, it is available again to store the “retline#” when “s1” is invoked. |
||
| 13 | post |
|
|||
| 14 | 2 | 12 |
| ||
| 15 | 3 | retline# | |||
| 16 | post |
|
|||