Buffer reuse gone wrong

[PJM_labview]

This bug is a little tricky to explain.

First download the attached VI.

Download File:post-121-1222274788.vi

Now Open its BD and from there open the FP of "Get Image Subset.vi" and put each FP side by side.

Step 1

Run the VI with the "White is transparent" flag set to FALSE.

Here is the result:

So far so Good (note: I highlight the structure where problem will show next)

Step 2

Run the VI with the "White is transparent" flag set to TRUE.

Here is the result:

Somehow the buffer of the cluster is being reused and some data that should not be there are present.

Note: at that stage there are no visible effect of this bug on the "Get Tree Symbols (LV86).vi".

Step 3

Run the VI with the "White is transparent" flag set to FALSE.

Here is the result:

Now we can see the effect of the bug on the "Get Tree Symbols (LV86).vi".

From that point on there are no non edit way to have this vi work properly. Setting the "White is transparent" flag to false always return an array of blank image symbols.

Disconnecting the image data wire (at any place) and reconnecting it does temporary fix the bug until going back to the 3 steps. forcing LV to not reuse the buffer by creating another wire branch (see image below) is the only workaround that I have found.

PJM

PS: This bug is also present in LV 8.5 (I have not tried earlier version)

[Mellroth]

QUOTE (PJM_labview @ Sep 24 2008, 07:17 PM)

...forcing LV to not reuse the buffer by creating another wire branch (see image below) is the only workaround that I have found...

I added a "always copy" primitive right after the Get Image property node, and this also solved the issue.

/J

[PJM_labview]

QUOTE (JFM @ Sep 24 2008, 10:36 AM)

I added a "always copy" primitive right after the Get Image property node, and this also solved the issue.

/J

Good point. I never had a use case for that little guy ("always copy" primitive). I guess this does qualify as a use case

PJM