[jawsscripts] Re: One of my wilder debugging discoveries
- From: Andrew Hart <ahart@xxxxxxxxxxxxx>
- To: jawsscripts@xxxxxxxxxxxxx
- Date: Thu, 21 Nov 2013 21:42:08 -0300
That's why I prefer to use all punctuation. Even though it's slower to
read, I make so many typos, etc., that it saves me time, I'm sure.
*smile* Well done trying that one down! I hope you didn't lose to much
time over it.
On 21/11/2013 1:27 PM, Doug Lee wrote:
> This just seems worth writing down for its oddity...
>
> I was debugging a 23,000-line script this morning. Recently I started
> getting really strange behavior when trying to assign collections to
> variables in part of the code: I started getting the integer 1 in
> place of a collection, with no obvious reason. I tried myriad
> solutions and debugging tricks, sayString/sayInteger lines, etc.
>
> Then I added a sayInteger at a particular point in the code, and presto,
> the thing started working! "How strange," I thought, "that shouldn't
> affect anything, it's not making an assignment..."
>
> So I took the new line out, and the odd wrong behavior resumed.
>
> On reading more carefully, I found that the next line after the
> sayInteger I had put in began with a spurious less-than sign (<).
> Apparently I had mistakenly copied that line out of a diff output
> file, in which less-than signs indicate removed lines and greater-than
> signs indicate added ones.
>
> I should mention here two things: I read with punctuation off, and in
> my actual code files, which are sometimes something like templates, I
> use a double less-than to indicate a translatable symbol; so I am
> accustomed to ignoreing less-than signs as I read anyway.
>
> The previous line in the file was an assignment like
>
> cLocation = locations[index]
>
> in other words, an assignment from an array to a collection variable. The
> line after that was a plain old function call, we'll say MyFunc(). So
> the code before my sayInteger was added looked like
>
> cLocation = locations[index]
> < myFunc()
>
> The JAWS script compiler treats end-of-line no different than a space
> most of the time, so it read the above two lines like
>
> cLocation = locations[index] < myFunc()
>
> which actually means, assign to cLocations the integer (boolean) value
> of the question, is locations[index] less than the return value of
> myFunc()?
>
> It so happens that myFunc() returns True, which as an integer is
> 1. I guess the integer value of the collection at locations[index]
> was 0, so the above question's result was True, which is also a 1
> when taken as an integer. This is how cLocation was getting set
> to 1 instead of a collection.
>
> Now, when I put the sayInteger call in, why did I get a different
> result? The code would then look like
>
> cLocation = locations[index]
> sayInteger(...)
> < myFunc()
>
> Now the spurious less-than sign doesn't affect the assignment. JAWS
> will actually make the comparison
>
> sayInteger(...) < myFunc()
>
> but then discard it because it's not being assigned to anything.
> This had the net effect of making my cLocation assignment work
> again.
>
> And they wonder why it is said we'll spend 90% of our time debugging. :-)
>
>
>
__________�
View the list's information and change your settings at
http://www.freelists.org/list/jawsscripts
Other related posts: