ArffExampleSource forgets decimal_point_character
Hi!
I'm having a strange problem. Sometimes ArffExampleSource seams to forget the value of its parameter decimal_point_character. I'm not exactly shure when this happens but I think it sometimes occures when i run a process and stop it without previously halting at a breakpoint. The parameter is then simply empty. When i restart the process i get an error.
I'm experiencing this error regularely since 4.4 (about once each two or three hours). But i have to admit that i haven't used Rapidminer 4.3 nearly as often.
Does anybody know what's happening? ???
Greetings,
Michael
I'm having a strange problem. Sometimes ArffExampleSource seams to forget the value of its parameter decimal_point_character. I'm not exactly shure when this happens but I think it sometimes occures when i run a process and stop it without previously halting at a breakpoint. The parameter is then simply empty. When i restart the process i get an error.
I'm experiencing this error regularely since 4.4 (about once each two or three hours). But i have to admit that i haven't used Rapidminer 4.3 nearly as often.
Does anybody know what's happening? ???
Greetings,
Michael
Find more posts tagged with
Sort by:
1 - 9 of
91
I also loose the decimal point, in my process in ExampleSource and in GuessValueTypes. It always happens after editing a parameter in some other operator. I tend to think it happens in particular when ending the editing with striking the <Enter> key, but I really did not detect consistent circumstances as yet.
Hi,
I'm still experiencing this error -- additionally with CSVExampleSource which I haven't used before.
I surely renamed some operators but not directly before the error occured. At least not every time.
When I find the time I will check out the source and run it in a debugger.
Greetings,
Michael
I'm still experiencing this error -- additionally with CSVExampleSource which I haven't used before.
I surely renamed some operators but not directly before the error occured. At least not every time.
When I find the time I will check out the source and run it in a debugger.
Greetings,
Michael
Hi again,
I finally managed to reproduce this behavior (at least partly). Here is what I did:
1. Add the operator ArffExampleSource (or one of the others containing a decimal point character) to an empty process
2. Change the decimal point character to two points
3. Add an empty operator chain
The complete process should now look like this one (please note the two point characters in the parameter):
Now select the chain and rename it (by pressing F2, a single click after some time after selection, or by contect menu - it doesn't matter). Select the example source operator and check the parameter: the points have been gone.
Although this is also a bug I can hardly believe that this is what happened to you guys since using double decimal points do not seem to make any sense at all. Can you confirm this? Any other thoughts?
Cheers,
Ingo
I finally managed to reproduce this behavior (at least partly). Here is what I did:
1. Add the operator ArffExampleSource (or one of the others containing a decimal point character) to an empty process
2. Change the decimal point character to two points
3. Add an empty operator chain
The complete process should now look like this one (please note the two point characters in the parameter):
<operator name="Root" class="Process" expanded="yes">
<operator name="ArffExampleSource" class="ArffExampleSource">
<parameter key="decimal_point_character" value=".."/>
</operator>
<operator name="OperatorChain" class="OperatorChain" expanded="yes">
</operator>
</operator>
Now select the chain and rename it (by pressing F2, a single click after some time after selection, or by contect menu - it doesn't matter). Select the example source operator and check the parameter: the points have been gone.
Although this is also a bug I can hardly believe that this is what happened to you guys since using double decimal points do not seem to make any sense at all. Can you confirm this? Any other thoughts?
Cheers,
Ingo
Hi again,
I can confirm this -- twice.
1. As I got an error message saying something like 'variable[0]: Array index out of bounds' I tried two points as decimal point separator. I thought this workaround would do but it didn't. Just as you showed.
2. As I've tried this workaround I'm absolutly shure that i previously used only one period.
Greetings,
Michael
I can confirm this -- twice.
1. As I got an error message saying something like 'variable[0]: Array index out of bounds' I tried two points as decimal point separator. I thought this workaround would do but it didn't. Just as you showed.
2. As I've tried this workaround I'm absolutly shure that i previously used only one period.
Greetings,
Michael
Hi,
Actually renaming operators is the only reason I can currently think of since we changed this for 4.4 a bit and periods play a large role in that. But since I tried at least 20 different renaming setups and a single period never got lost I am hoping that this could be the reason.
Cheers,
Ingo
Ok, but maybe this one did not get lost and this only happened for renamings after changing it to two? (at least I am hoping this).
2. As I've tried this workaround I'm absolutly shure that i previously used only one period.
Actually renaming operators is the only reason I can currently think of since we changed this for 4.4 a bit and periods play a large role in that. But since I tried at least 20 different renaming setups and a single period never got lost I am hoping that this could be the reason.
Cheers,
Ingo
Hi at all,
just to give you a notification: the decimal point loosing bug is fixed. The bugfix is already part of the recently issued RapidMiner Enterprise Edition update and will of course also be part in the next RapidMiner Community release which is about to come in a couple of weeks.
Kind regards,
Tobias
just to give you a notification: the decimal point loosing bug is fixed. The bugfix is already part of the recently issued RapidMiner Enterprise Edition update and will of course also be part in the next RapidMiner Community release which is about to come in a couple of weeks.
Kind regards,
Tobias
hmmh, we actually do not have had anything like this before. I just have tried different stress tests and tried to reproduce this behavior but without success. So if you are able to narrow this one down and have any additional idea why this might happen please let us know.
Cheers,
Ingo