SMAT DB resend to file doesn’t respect NOT

[Mike Strout]

I just thought I would post heads up to the group about a serious bug (that will be reported to Infor).

I was trying to filter and send a collection of inbound lab orders that contained “Path,Cyt” and didn’t contain “|SURG^”.

I used the advanced criteria because I don’t trust the smart search. In the first criteria I put the “Path,Cyt”. In the second, I checked the Not box and then entered the “|SURG^” criteria.

I clicked the search icon and got 16 results back. I then clicked the header checkbox to select all 16 records and then resent them to a file. The dialog box even said “16 messages resent to file”. To my surprise, 37 messages were dumped into the target file and even worse, every one had |SURG^ in PV1-3.

This new development has me worried about every message resend I have done since we upgraded to 6.1.2.0.

Update: I put on my programmer’s hat and wondered if the developers of this function only looked at the Not in the first criteria, so I thought I would try flipping the order from my search above. I put Not |SURG^ in the first advanced criteria and “Path,Cyt” in the second. I received the same 16 messages, so I tried to export them to file. I say “tried” because this time the resend to file sent 0 messages (even though the dialog box said it resent 16), so no file was created.

I decided to remove the “Path,Cyt” criteria and only send the first 200 messages where “|SURG^” is Not in the message. Same results. If there is a not in the first criteria, no messages will be resent to the target location.

I hope Infor already knows about these and has fixed them in the hotfix we are about to implement.

Topic

[Charlie Bursell]

Mike:

Try a negative lookahead.  I found out that the regexp engine in CL is Per based and Perl uses the same lookahead as Tcl

Your regexp would look like:  “(?!|SURG^)Path,Cyt”

The ?! says to lookahead and only match if that expression not found.

I have not tested it but worth a try.  I will test it later in my Linux system.  For some reason SMATDB does not like Windows 10 32 bit  😀

FWIW, I did try it with my hcismatdb script and it did work there

[Mike Strout]

Thanks Charlie. I am pretty disappointed with my 6.1 upgrade. I had such high hopes for the SMATDB, but it seems every time I turn around, I find another bug that makes searching it untrustworthy. Something as simple as…

A|B|C|D|E|F|G|H|I|J|K

…will fail because it has more than 10 options. If it can’t handle something as simple as this, I have a hard time believing it can handle more complex regular expressions like a negative lookahead.

[Charlie Bursell]

Mike:

I did test it using SMATDB and could not get the negative lookahead to work.  Perhaps you should contact Support and write a bug or enhancement request.

As I said, for a temporary work around you could use my hcismatdb script.  I have not tested resend to a thread thoroughly but I have tested it a lot sending to a file.

Let me know if I can help.

[Tim Zwieg]

Mike

Did you submit this to Infor as an incident?  Similar issues with Smat DB myself.  I searched Smat using advanced time range, there were 70k total messages in Smat.  My advanced time range showed 19k messages.  I used the Mark All messages button, and did a resend to file.  My file contained 70k messages even though it said 19k messages were selected and resent.  However in another search of the same messages, if I used the checkbox to select all messages then I got the 19k as expected in my file.

[Mike Strout]

Yes, I did open a ticket that has been escalated to development. During my discussions with them, they said that the GUI and the resend process use different regex libraries, which accounts for the differences what is displayed vs what is resent.

[Tim Zwieg]

Thanks Mike, I too opened an incident with them, after doing some testing this AM.  I will  post the text of what I sent to Infor.

[Author]

Replies