"User-defined Macros not resolved for MetaData checking"
cthiel
New Altair Community Member
Problem description:
I define a Macro using the "Set Macro" operator.
The next operator uses this macro.
It will work correctly.
However, the process checker will complain, showing a Problem-Message that indicates that for the process-checking/metadata-generation the String %{MacroName} is simply replaced by MacroName and not the content of the macro.
An example process for this is available in bug 367, http://bugs.rapid-i.com/show_bug.cgi?id=367
The problem also persists in SQL-queries, see bug 377, http://bugs.rapid-i.com/show_bug.cgi?id=377
I find this problem very confusing. More, I switched to Sun JDK, and the error disappeared, but only for some hous. A clean install on another machine (now Windows, not Linux) and recreating the test process by hand had the same problem. Dito for a third machine. I am very confused as to how this bug could ever have disappeared.
In summary, the macro-replacement does work when the process is executed, but not when the process is checked. Which destroys the metadata, leading to 20+ warnings for a medium size process.
Does anyone have ideas what the problem might be? And can somebody try to reproduce it? (Despite it being tested on 3 machines, 2 OS...)
Thank you,
Christian
I define a Macro using the "Set Macro" operator.
The next operator uses this macro.
It will work correctly.
However, the process checker will complain, showing a Problem-Message that indicates that for the process-checking/metadata-generation the String %{MacroName} is simply replaced by MacroName and not the content of the macro.
An example process for this is available in bug 367, http://bugs.rapid-i.com/show_bug.cgi?id=367
The problem also persists in SQL-queries, see bug 377, http://bugs.rapid-i.com/show_bug.cgi?id=377
I find this problem very confusing. More, I switched to Sun JDK, and the error disappeared, but only for some hous. A clean install on another machine (now Windows, not Linux) and recreating the test process by hand had the same problem. Dito for a third machine. I am very confused as to how this bug could ever have disappeared.
In summary, the macro-replacement does work when the process is executed, but not when the process is checked. Which destroys the metadata, leading to 20+ warnings for a medium size process.
Does anyone have ideas what the problem might be? And can somebody try to reproduce it? (Despite it being tested on 3 machines, 2 OS...)
Thank you,
Christian
0
Answers
-
Hi there,
Are you on a commercial license? The reason I ask is that your email was showing as from http://www.crealytics.de/ , so I assumed you're on some OEM deal.
0 -
Hi Haddock,
no, still on a community licence. I use it at work, hence the reference.
In how far is this relevant?
Greetings,
Christian0 -
Hi,
well it could be relevant for the question if and how fast your feature request (see http://rapid-i.com/rapidforum/index.php/topic,1704.msg9154.html#msg9154 ) would be fulfilled
Cheers,
Ingo0 -
Greets Christian,
Ingo got it right about why I asked whether you were on the commercial edition, but let me expand..
Recently the forum has welcomed some exceptional new members, data-mining Ph.Ds and such, like yourself; it has also welcomed lesser mortals, and things came to a head recently...
http://rapid-i.com/rapidforum/index.php/topic,1731.msg9069.html#msg9069
You will see from that thread that I raise the issue of lesser services for those that can basically look after themselves, but might need occasional help. Given that you presumably make money out of your use of Rapidminer I thought you might be a case in point. Was I wrong?
0
