Yes, I am using the dreaded "B" word here. Many of these are bugs. I am hoping to put all of these in a single place available to the
info-labview community. I have found that there are a large number of errors that have gotten thru QA on 6.0.2. These are just the sample
that I have reported or heard on Info-Labview and verified most of these have been verified by NI and are awaiting fixes. There should be a
lot of patches for version 6.1.0!
NI issues a CAR (Corrective Action Request) on these if they are reported to tech support but it is unclear when and if any action is taken.
I know from personal experience some issues are closed by developers without any fixes.
Please keep sending them to the list and if you have an example VI send it to me and I will try to collect these together. If you can
verify the problem or that it does not occur on other platforms I will try to incorporate it. If I don't list a platform then I have no
data.
Unfortunately this has gotten too big to send to the list since emails over a certain size are banned to protect the bandwidth limited.
-Scott
- Interrupt Not Available - Scott Hannahs <sth@magnet.fsu.edu>
Date Reported: 2/6/2001
NI-DAQ 6.6.0f3 (Mac) aborts after 31162 read iterations on PCI-MIO-16XE cards. Gives PeeCee type error about interrupt Level not
available.
Example VI
Platforms: Mac - Fixed in NI-DAQ v 6.6.1 download
Corrected Platforms: W2000
- The VISA documentation files are corrupted - Scott Hannahs <sth@magnet.fsu.edu>
Date Reported: 2/23/2001
Look at the documentation for a VISA function (VISA Clear I think). I have also seen the problem with VISA STB, VISA WRITE, VISA Open
And VISA Close. As you can see, right after the label "error in" everything goes south. The "no error" should be in parenthesis, then
it has a bunch of stuff then the "dup VISA resource name" is on the same line. The error out description is messed up as well.
Platforms: Mac - Status to be fixed by conversion of Help to HTML someday
Corrected Platforms: W2000
- Documenation Does not open DAQ examples - Scott Hannahs <sth@magnet.fsu.edu>
Date Reported: 2/23/2001
Open LV 6.0.0 (or 6.0.2). Select the menu Help->Examples and QuickHelp opens with a list of topics for examples. I chose "I/O
Interfaces" and then "Data Acquisition". At that point I get a simple message that I must have NI-DAQ 6.6.0 installed to access the
examples. The problem is that I DO HAVE Ni-DAQ 6.6.0 installed.
Platforms: Mac - Status to be fixed by conversion of Help to HTML someday
Corrected Platforms: W2000
- LabVIEW failure on Line 815 of grabber.cpp - Scott Hannahs <sth@magnet.fsu.edu>
Date Reported: 2/23/2001
- Start LV and open new VI
- Place a "meter" on the FP
- Switch to selector tool (arrow)
- Select meter
- Menu Edit-Customize
- Grab bottom of middle of the Green→Red shaded area (the bottom edge below the "5.0" value, and drag this down a ways
to widen the shaded region
- Switch tool to tweezers "customize"
- Move cursor to lower right of the meter and it should change to the symbol for the ability to expand a box. This cursor
looks like a couple of nested 90 degree angles.
- Click the mouse as if you were going to expand the area. Boom. Get the grabber.cpp error
Example VI
Platforms: Mac - Corrective Action Request filed
Corrected Platforms: W2000
- Clipboard Error
If copy an array of booleans to the clipboard and paste into another application (ie GraphicConverter) it is badly corrupted. It
appears that the transparent background is not copied to the clipboard or cleared. This happens on many controls.
Platforms: Mac, WNT
Corrected Platforms: W2000
- Drop thru clicks from other applications cannot be disabled.
Platforms: Mac - Corrective Action Request filed
Corrected Platforms: W2000
- Error in "Extract Numbers" example.
Reported by: Scott Hannahs <sth@magnet.fsu.edu>
Date Reported: 2/1/2001
Extract Numbers in the Examples/General/Strings.llb is an extremely useful function. In LV 4.1.1 this worked. In LV 5 some genius
hardwired the delimiter to be a comma. I reported this problem with a plea to go back to the 4.1.1 version. In LV 6 they attempted a
fix but didn't think too hard about it, now the VI can hang in an infinite loop. To check this out, enter the string "--5" in the front
panel and run it. It will never exit. PLEASE GO BACK TO THE VERSION 4.1 ROUTINE IT WORKED JUST FINE!
On further inspection there was a flaw in the Version 4.1 routine. I have improved it so it now will work in all cases. I think I have
gone back to something similar to the version in LV 3 or so. But anyway this is a simple useful utility that they seem to have trouble
with. I am sure this is cross platform.
Fixed Example VI
Platforms: All, to be corrected in the next release
- Memory Leak in create/destroy notifiers (not to be fixed until V7??)
About every 2500 create/Destroy cycles it allocates another 64K pointer. And about every 10K create/Destroy cycles it allocates another
32K handle. This is an unusual circumstance, to do this so many times, but for long running programs it can crash after months etc. I
talked to NI and they are aware of it and are considering fixing it for LV Version 7 when they do a complete rewrite of the
synchronization VIs. This is too long to wait for a bug fix.
Example VI
Platforms: Mac, W2000
- Color gradient disappears on large meter displays.
This is due to the shareware rendering engine LV uses on the Mac.
Example VI
Platforms: Mac
Corrected Platforms: W2000
- Digital waveform graph crash
A crash bug appears in the digital waveform graph since labview 6.01 We found this bug in the 6.02 fix but later research showed it was
a bug already since 6.01. You can reproduce this bug as follows:
- put the digital waveform graph on your vi
- extend the legend upwards (downwards does not show the bug)
- change the color of the graph
- boom crassssssh due to incorrect memory addressing.
Platforms: W2000
Corrected Platforms: Mac
- Cursor Plot Property "Craig Graham" <c.n.graham@lancaster.ac.uk>
Date: Mon, 26 Feb 2001 20:22:32 -0000
Don't wire a negative signed int to the unsigned input of the "Cursor.Plot" property of an XY graph otherwise you get a crash in
"cursors.cpp" when that portion of code executes.
- Create XY Graph
- Show cursors and place one cursor in middle of graph
- Create property node change to "write" and create a control
- Change representation of control to I32
- Enter -1 into control and run VI.
Get Failure at graphcrs.cpp (line 2360 MacOS9, LV 602. line 2366, Pmac G3 MacOS8.6 LV6.0.2) When you fill the graph with data and
select cursor etc by property node using correct values, run it, alter the representation of the control to I32 run it and select value
-1 one : NO error !!.........most times :^))
Last time I did LV crashed due to error 2 ......
Example VI
Platforms: Mac, Win
- Variant bug Dominic Lavoie <DLavoie@itfoptical.com>
Date: Mon, 26 Feb 2001 11:23:54 -0500
When you transfer value from a cluster to another cluster, you give new name to each element. If you use the variant to flattened string
function, you can see the new name in the type descriptor. The problem occurs when you make twice similar unbundle in the same diagram
and that even if both part have no connection between them. The problem is that the second variant to flattened string gives the type
descriptor of the first one. But that problem occurs only if you make an implicit conversion. When you make an explicit conversion,
everything is ok.
- Units bug
Goal: Display temperature with degC and degF using one control
- create a thermometer indicator; show the digital display
- edit the digital display to replace it with a second thermometer
- add units(degC) to the first indicator.
- BUG: the second indicator has DegC units on its label but it displays Kelvin(temp. base unit) values on the scale. LV accepts to
change the units of the second indicator to degF but values are not yet consistent with the first.
Example VI
Platforms: Mac, Win
- Table Bug - Robert C Burton <burtonrobert@excite.com>
Date: Wed, 6 Dec 2000 22:46:01 -0800 (PST)
I discovered that if you create a table in 5.1.1 then open it in 6i and try to change the table index you get an error: "support.cpp"
line 7182. (line 7240 in 6.0.2) I sent a report to NI but have not received a reply yet
- Open LV 5.1.1, New VI, Place a Table and Save
- Open this VI in LV 6.0.2
- Highlight an index and type a new number.
- Hit enter.... LV Failure.
Example VI
Platforms: Mac, Win
- VISA problems Uwe Frenz <uwe.frenz@getemed.de>
Tue, 27 Feb 2001 09:57:50 +0100
For instance, VISA does not work, even the examples provided do not. For the German version one needs a replacement for some files to
get it ready.
- Localization problems (German) Uwe Frenz <uwe.frenz@getemed.de>
Tue, 27 Feb 2001 09:57:50 +0100 There was no located version of 6.0.1 and the located 6.0.2 was later than the English original version.
The located help system is even more (much more) buggy than the English one - content just copied, explanations left out, variables
explained, that don't show up on that particular function ... I am missing an explanation of the new server functions and the nested
reference model, what is available where.
- Graph Printing "Paul Rijkers - (ieniemienie)" <scanonym@IRI.TUDELFT.NL>
Date: Tue, 27 Feb 2001 13:51:03 +0100
Try to do a printout of a CP with an Ygraph in it, with the thickest linewidth to a postscript colorprinter. The legend shows a nice
thick line. The graph itself however shows a very thin line! This was an old bug (I think it was lv4.?), was gone in LV5x and now back
again! The same in W2000 on a color and a black and white postscript printer. But surprise surprise.. Standard printing also gave a thin
line, even in the legend.
Platfroms: Mac, W2000
- Real Time Stamp Counter "Maillard Patrick" <Patrick.Maillard@ch.bosch.com>
Date: Tue, 27 Feb 2001 14:41:46 +0100
Relative to a thread that has been on the list some time ago, I looked at the rdtsc.vi
that is in the NI developer zone and I found a bug in a subvi (wrong calculus of Diff L in 'RDTSC Diff To Time.vi')
Therefore, I pressed on "Add your comment" at the bottom of the page to signal it. The answer I received by email from NI was quite
disappointing: "It appears that you are correct; however, since the "DIFF L" and DIFF H" indicators are not used downstream in the VI,
it's somewhat of a moot point. The subtraction can easily be corrected by rewiring to obtain "DIFF L" from "2L" - "L". Again, since
"DIFF L" is never used, this doesn't really even matter."
Platforms: Win (All)
Corrected Platforms: Mac, Unix (not applicable, uses CIN)
- String to IP bug Christopher Burian <cburian@ll.mit.edu>
The examples in udpex.llb demonstrated that UDP broadcasting does, in fact, work. The bug is in the String to IP function, which returns
0 instead of 0xFFFFFFFF for an input string of 255.255.255.255.
Example VI
Platforms: Win NT4 (LV 5.1), Win98 (LV 5.1)
Platforms Corrected: Mac
- Application Builder Crash "Kniskern, Roy" <KniskeLJ@litton-ps.com>
I had Application Builder crash when building one of my Apps in 6.0.2 (It works fine in 6.0). Reported to and reproduced by NI. It
occurs, apparently, when the "Do not compress target file option" is selected if the application uses NiDAQ vi's.
- Multicolumn Listbox - Cat <cat@catscreek.com>
Date: Tue, 27 Feb 2001 09:05:53 -0500
The multicolumn listbox does not scroll automatically as the selection cursor is taken past the bottom (or the top) of the control, even
tho the actual value of the control is updated.
Drop a multicolumn listbox control on a front panel. Fill up 30 or so rows. Resize the multicolumn listbox so you only see 5 or so
rows (keep the scrollbars on). Make sure there is lots of empty front panel space below the control. Set the control to 1 item
selection (it really doesn't matter what it's set to, but this makes the example easier). Next to it, create a 1D i32 array indicator.
Wire the two together. Wrap them in a while loop. Run the program.
Put the cursor on the top row and press the left mouse button. Slowly drag the cursor down, keeping the button pressed. Each row is
highlighted in turn, as would be expected. In the array next to the control, you can see which row is currently selected.
Now comes the fun part. Continue to slowly drag the cursor down until it goes past the bottom of the control. The multicolumn listbox
display does not change. But watch the array. Even tho the cursor is completely off the control, the actual value of the
multicolumn listbox selection keeps changing as the cursor moves, relative to the row it would be on, if the control were
actually there.
What it should do:
Repeat the above construction, only use a plain ole listbox control and scalar i32. As the cursor is dragged down to just below the
bottom of the listbox, the listbox begins to scroll and shows the row corresponding to the current value.
Example VI
Platforms: WinNT, Win98, Mac
- Printing of Anti-Aliased Plots - Jose Olcese <labview@tutopia.com>
Create a waveform graph, select antialasing to plot 0, then write some data to the graph (100 random numbers) and finally try to print
the front panel..... NO data printed on plot
Example VI
Platforms: WinXX, Mac
- Grid Lines on Chart scroll away - Urs Lauterburg <urs.lauterburg@phim.unibe.ch>
Date: Sun, 04 Mar 2001 22:47:31 +0100
Chart vertical grid lines do not seem to update once the drawing area starts to scroll. This screws up all my very many scrolling charts
when taking those pieces of code from LV 5.1 to 6.0.X. Is this the reason why charts would newly come up without grid lines when
dropped to the FP? Is there a trick to make vertical grid lines scroll as opposed to just vanish. I don't like this at all. Happens on
both new 3D an "Classic" Waveform Charts.
Example VI
Platforms: Win2000, Mac
- Multicolum listbox in created subvi - Cat < cat@catscreek.com>
Date: Tue, 08 May 2001 10:21:12 -0400
LabVIEW crashes (failure in prop.cpp line 2210) when attempting to "Create Sub-VI" with a multicolumn listbox local variable in it.
This is repeatable.
Platforms: WinNT, Win98, Win2K, Mac
- Templates not renamed correctly - Jürg Bögli <juerg_boegli@hotmail.com>
Date: Thu, 10 May 2001 10:54:19 +0200
I have made a VI template and saved it as .vit. Now when I open it, it still keeps the original name and type like "Name.vit". This
bug appeared with LV6. I already sent a mail to NI last October. The same goes for .ctt controls.
Meanwhile you can use this workaround (suggested by NI):
"The behavior can be avoided by using the Select a VI palette function to place it on a diagram as it loads a copy of the template."
Platforms: WinNT, Win98, Win2K, Mac
- PID toolkit bad when proportional only control - Bruce Ammons <bruce@ammonsengineering.com>
Date: Thu, 10 May 2001 11:54:27 -0400
I have found a few bugs in the toolkit. They showed up when I tried to do proportional only control. I found that the integral portion
(which should always be zero for Kp only) would get set to a constant offset. NI has been notified, but I don't know if they have done
anything about it.
The problem is most noticeable when you exceed your limits. Set the proportional gain to a large value, and use a setpoint with a large
error. The limits will be exceeded and the integral constant is adjusted to stay within the limits. Now, if you set the setpoint equal
to the error, the output should be zero, but it isn't. It is still equal to whatever was used to compensate for the limit. This value
is stored in a shift register in the integral routine. I ended up modifying it to always use zero, except to correct for exceeded
limits.
NI response: "I spoke the the Research & Development Engineer that will be dealing with your request for a change to the PID toolkit
vi's. He was aware that for some applications this behavior was not expected nor wanted, and feels that he can change the behavior for
your. I do not have a time estimate at this time, but the corrective action request is CAR 29UG1MBQ"
Platforms: WinNT, Win98, Win2K, Mac
No example posted since PID toolkit is not public
- VISA Find does not return a VISA resource - Scott Hannahs <sth@magnet.fsu.edu>
Date Reported: 4/2001
The output from VISA Find resource is not compatible with the VISA Open function. Need to type cast the output to a VISA resource.
Note sent to developers.
NI response: "VISA Find will continue to return an array of strings. In the next version of LabVIEW, the other VISA functions should be
able to accept a string without requiring a cast. It will arbitrarily pick a valid matching resource class, such as the generic
"Instr", so if you want something more specific you would still need to cast. But this should work for almost all users."
Platforms: Mac
- VISA Serial Set Buffer Size does not work - Scott Hannahs <sth@magnet.fsu.edu>
Date Reported: 4/2001
The set serial buffer size does not set a buffer size greater than 31 bytes.
Example VI
Platforms: Mac, corrected by patch. Windows
patch stops LV from being a CPU hog while waiting for input.
- VISA Serial Set Buffer Size returns warning - Scott Hannahs <sth@magnet.fsu.edu>
Date Reported: 4/2001
The VISA call to set a serial buffer size returns a warning that it cannot set the buffer size. However it appears that the buffer size
is set correctly. This occurs after the patch above is applied. Mac does not support setting output buffer size, documentation needs
updating.
Example VI
Platforms: Mac
- VISA Serial Flush Buffer does not work - Scott Hannahs <sth@magnet.fsu.edu>
Date Reported: 5/2001
The VISA flush receive buffer does not clear the receive buffer.
NI response: "You are correct! This is a new CAR 29TCS14U. We will try to get that fixed in the next version of LabVIEW, but no
guarantees at this time. At least you have a valid workaround by querying the number of bytes present and then reading those bytes and
throwing them away."
Example VI
Platforms: Mac
- Crash when uncheck "Show Menubar" A hard LabVIEW crash with VI's that are set to ''Size the Front Panel to the Width and Height of
the Entire Screen'' with ''Window has Title Bar'', ''Show menu Bar'', ''Show Scroll Bar'' and ''Show Toolbar When Running'' all
unchecked. It happens when the cursor arrow is moved to the Macintosh menu bar area and when clicking on it....wham it goes and the
system recommends a new start.
Info from Bien Entendu <BienEntendu@mac.com>
it just happens when you uncheck "show menubar" there is a conflict with the STF Fax software. so you have three options :
- don't use the fax software by disabling all the fax stf extensions in extension manager
- set show menubar to "on" and set a menu for every vi with a frontmost panel
- convince national to repair this bug or to correct the bug on "show menubar" which still shows a menubar when the option is
unchecked(!) and by the way, ask them to remove this ridiculous and disgratious white bar on top of every VI(where the menubars are
on the other platform)
Example VI
Platforms: Mac
- VISA parallelism bug on lock resource async - Scott Hannahs <sth@magnet.fsu.edu>
Date Reported: 4/2001
Under a race condition, the lock resource could incorrectly return a timeout. NI Response: "The invalid VISA lock timeout bug, CAR
27FE3UIL, has been fixed in the next version of LabVIEW".
Platforms: All
- Crash in "typemutation.cpp" when opening VI - Socratis Kalogrianitis <Socratis@imagetherm.com>
Date: Wed, 06 Jun 2001 15:50:00 -0400
Take a gauge control. Visible items→digital display. Select just the digital display and Edit→Customize control. Replace it
with a text ring (and fill in a couple of values, optional). Close the editor. Save the VI with the modified control. Now try to re-open
that VI. Kaboom...
Failure in "typemutation.cpp", line 361
It happens with all numeric controls that have a digital display. The biggest harm is that the problem shows up AFTER you have saved the
VI and you try to reopen it. You better have a backup...
Also, if you try to change the change the Representation (to I32 let's say) you get another crash in "proptyp3.cpp"...
Platforms: All Win and Mac
- GPIB Hang - Scott Hannahs <sth@magnet.fsu.edu>
Date Reported: 7/25/2001
There is a problem with asynchronous GPIB under LV 6.0.2. This is using VISA but I believe it is basic to the GPIB driver. If one does
an asynchronous GPIB read and it times out, subsequent GPIB reads do not timeout but hang forever.
- Device on bus at address 1
- Do async read on address 15, will timeout.
- Do async write, and aysnc read again on address 15
This second asynchronous read will not timeout but hang until the abort button is pressed. Memory is corrupted since if I quit LV at
this point, I will get a crash into Macsbug.
Example VI
Platforms: Mac
- shift register bug - Rick Beauchaine <rjoebeau@hotmail.com>
Create a vi with a while loop and shift register. Initialize the s/r with an error cluster (constant or otherwise). Insert two bundle by
name nodes (set to 'Status') in series with the error cluster. Wire a False boolean constant to the first and a True to the second (or
the other way around). Insert an unbundle by name node at the output of each bundle node. Wire the boolean Status output of these to the
edge of the loop and create 2 indicators OUTSIDE the loop. The loop does not have to iterate more than once so no front panel button is
necessary, just wire a bool constant to the condition terminal.
The boolean attached to the first unbundle will follow the value of the second boolean when the loop terminates, even though a probe or
indicator in the loop shows otherwise. The first boolean is being changed at the while loop output tunnel!
Example VI
Platforms: Mac, Win NT, Win 2K
- Insane Object from Typedef - Paul F. Sullivan <Paul@SULLutions.com>
Date Reported: Tue, 4 Feb 2003 10:45:53 -0500
The attached library has a very simple demonstration of the insane object problem I have been working with. It fails in LabVIEW 6.02 on
Mac and Windows 2000 (the environment of my client). Having built the simple demo, this morning I checked it under 6.1, which does not
show the problem. To see the problem, open the two VIs and the type def in the library in 6.02. The VIs are almost identical, except
for the cluster constant being set to Autosize→Arrange Horizontally (which gave a much better fit in the diagram of my original VI)
in Goes Insane.vi. Both of the VIs should run fine. Edit the type def, e.g. by changing "Wait For" to "What For". Close the type def to
propagate the change. (Refusing to save it makes it easier to repeat the experiment.) Touch the run arrow on either VI and see the
report of insane objects in Goes Insane.vi but not in the nearly identical Simple Script.vi.
Example VI
Platforms: Win 2K, Mac
- Phantom Label - "Paul F. Sullivan" <Paul@SULLutions.com>
Date Reported: Tue, 8 Apr 2003 08:57:06 -0400
Using a property node to read the visibility of the caption of a control that has never had the caption visible will move the label to
the alternate position and leave a phantom copy of the label in the original position. This phantom is not attached to the control and
will not be redrawn if any object is dragged through any part of it. The problem is corrected in 6.1.
Platforms: Mac, Win 2k
- No Decoration Refnums - "Paul F. Sullivan" <Paul@SULLutions.com>
Date Reported: Wed, 23 Apr 2003 11:21:00 -0400
The properties All Objects[] and Selection List[] return Not a Refnum for front panel decorations. This problem is fixed in 6.1.
Paltforms: Mac, Win 2k