PREV INDEX NEXT

Author: Jau
Subject: *IMPORTANT*: Bug in the reference solution for BashLT
Date: Wednesday, 02 Dec 2020, 17:41:52

Hi all,

There is a bug in the reference solution for bashLT, as some of you have realized and pointed out in this newsgroup.

Right now, the reference solution exhibits the following behavior:

>  (1)$ export VAR=VAL
>  (2)$ printenv VAR
>  (3)$

In other words, the value of the exported variable VAR does *not* persist beyond a single call to process(). This is, however, *incorrect*, and the exported variable VAR *should* persist while the same parent shell is running. The expected behavior should thus really be:

>  (1)$ export VAR=VAL
>  (2)$ printenv VAR
>  VAL
>  (3)$

You can use /c/cs323/Hwk5/bashLT.V0 to examine this behavior (instead of using /c/cs323/Hwk5/bashLT). In fact, you can use /c/cs323/Hwk5/bashLT.V0 to check expected behavior in general, except for possibly memory leaks (the change from /c/cs323/Hwk5/bashLT.V0 to /c/cs323/Hwk5/bashLT was made to fix memory leaks).

TO BE EXPLICIT, YOUR CODE NEEDS TO FOLLOW THE LATTER BEHAVIOR RATHER THAN THE FORMER, EVEN THOUGH /c/cs323/Hwk5/bashLT CURRENTLY FOLLOWS THE FORMER.

Note also that /c/cs323/Hwk5/bashLT is not expected to change anytime soon (or maybe at all), because, as you all know, Prof. Eisenstat is currently in the hospital (and we are all hoping for a speedy recovery for him).

There is more discussion of this on the spreadsheet of answers at https://docs.google.com/spreadsheets/d/16wF_tHLnXbvW9z0i18XVED-Id8t3cfOBTKP8QrL6nqQ/edit#gid=0, if you have not viewed it already.

Thank you all!
PREV INDEX NEXT