Problem in combining feeds while using operator "aggregation of column"

Doris,

Your explanation is very detailed, but I would appreciate a screenshot a lot. Could you provide a screenshot of the feed definition as well as screenshots of the debug view of both incoming paths of the JOIN operator.

Thanks.

Thank you, Markus, for your reply.

Inserting images via ARIS Community browser is a pain in the a..., sorry. As I do not want to spend more time cropping and resizing images (every time I have changed the image I get a new message concerning restrictions of the size, on time a maximum in kb, after that the maximum witdth in pixel,...) I will send them to you via eMail.

I would like to remark something: In fact the SAP table contains a date for the update of data and a field for the update time but as the latter shows content that differs from the SAP time stamp I chose to use the timestamp to rely on information of date and time.

Thank you for your support,

regards,

Doris.

Yes, uploading images is not easy and time consuming.

Assuming you have MS Office on your machine: It ships with an application that eases image preperation for upload somehow. It is called "Micrsoft Picture Manager". Although Office itself allows you to crop and resize office hides the actual pixel sizes and shows cm/inches instead.

The Picture Manager reveals pixel sizes which are relevant for Community purposes. Once each side of your image is smaller than 700px and it is JPEG compressed upload should be fine. The image below shows edit image, resize options.

Of course every other image authoring software like the open Gimp was fine, too.

Hi Stephan,

thank you for your reply, too.

Actually I used Microsoft Picture Manager to resize the images. Unfortunately the system returned different notes concerning the size. During the first upload it told me the size in kb was too large. I changed that for the images to fall below that kb size but after the second try the system told me the width in pixels was too large. Every time it used a new measurement to tell me the image size was too large. After the second try (which took some time as I wanted to crop several different images) I didn't feel like having a third (fourth,...) try.

Probably the way of uploading images should be revised. It is not very motivating to spend so much time on that not being successful anyway. Probably the required size (in all of the n measurements :-)) should be provided in advance and the tool should be recommended, or the restrictions in size could be changed...

Regards,

Doris.

We just deployed a new version of the post/comment editor. The new editor comes with an updated image handling capability. Instead of rejecting an image, which is too large, it will scale down the image automatically. So this should solve your problem mentioned above.

Still, I advise to do the resizing and cropping manually, because it gives you full control of how it will look in the article. Especially fonts get unreadable easily and some experimenting is needed to find good scaling factors. In general, images are limited to 700 pixels for each dimension.

Thank you, Sebastian, for your information. To me this sounds like a very good solution for the "image issue".

Regards,

Doris.

Converting a SAP timestamp

In many SAP table the timestamp is stored in a representation like this "20100511144421", which means 5 May 2010 14h 44min 21sec.

If this pattern is used in a CSV file, MashZone tries to detect the correct data type and of course this could also be interpreted as number (this is the auto-detect data type feature). In this case, it is simpler not to convert into a number format (with the possibility to loose precision), but to keep the text representation. This could be achieved by configuring columns in the data source operator.

The type column contains the auto-detected data type, which can be overwritten by the checkbox in "Type as text" column.

Next the "Change data type" operator could be used for converting the text into a correct data format. The pattern for this operation could be specified. In this case it is yyyyMMddHHmmss, where yyyy is the year portion, MM the month and so on. Please note the case-sensitive letters.

Finding the latest timestamp

Doris,

I hope I understood the problem correct. From a list of time stamps, you need to identify the most recent. That can be done via the "Aggregate" operator. For this, the timestamp has to be converted into a number, because only a number can be used as an aggregation column. The conversion from type "date" into "number" in Mashzone results in something like milliseconds since 1 Jan 1970, so bigger numbers mean more recent timestamps. Identifying the maximum value on this number, is latest timestamp.

So, in my solution the left-hand path is determining the most recent entries per "Data target" in the table. To merge this together with "Duration" and the "Updated" timestamp in a readable format, I joined both tables together with two criterias as keys: the "Data target" of course and a second key, which is the number-converted timestamp.

Hi Markus,

thank you again for your suggestion. please excuse my late reply.

Unfortunately, the feed you suggested only works as "standalone" solution. The problem is that I do need this information in a larger feed which consists of various steps, transformations, etc. I have tried to adapt my large feed but it seems that sooner or later I get to the point where I cannot combine the various sequences any more.

To get the duration as an intermediate result -which is not contained in the original table- I have to perform an arithmetic operation which subtracts the time of data request from update time. In another small sequence I aggregate the sap update time stamp with "maximum". I can then combine these two intermediate results to get one that contains data target, latest update time and duration of this update.

The problem is that, to integrate this intermediate result into the rest of the large feed, I would need a source like yours, with data target, updated (maximum) AND duration. Otherwise I get the same notification "more than one column..." when trying to combine the intermediate result "data target / duration / last update time" with the rest of the large feed.

I guess I would have to try to integrate the duration into the second sequence to combine the result of the two mentioned sequences to my intermediate result (which has to be combined with the rest of the feed then).

As it gets confusing now I would like to suggest to check this in a personal conversation rather than trying to describe the various steps in my large feed here.

Regards,

Doris.