Hi
I have been doing some development with PAW Modeler recently. Overall, it is a good product and it is getting close to being able to do what Architect can do on the Web. It brings benefits such as colour coding and Ctrl-Space lookup of possible parameters, functions, Find and Replace, etc. You can now float a window out and drag it to a second monitor.
The biggest deficiency for me is that when you are developing a process, you are likely to get errors and there does not appear to be anyway to access the TM1ProcessError log file from PAW Modeler, unless I have missed something. Just to make it clearer we frequently have processes that call other processes, etc, so we need to be able to access all the TM1ProcessError logs as the error may have originated higher up. In order to debug you also need to be able to write out text files and be able to access those. Ideally, just giving file access to the Logs directory whether on Cloud or On-Premises would do it. It is this deficiency that has me RDPing onto the server and reverting to Architect when I need to debug a process.
All the other points for improvement are relatively minor.
Of new things in Modeler that need a little fine tuning.When you want to create a object such as a Cube, Dimension, Process, etc, instead of needing to scroll to find Processes and right click on that, you can now just click eg Process at the top. The problem that arises is that if more than one TM1 Server is listed, you then get prompted to select the server on which to create the object. If there is a long list of servers it can take as much time to scroll through the list as it does to scroll to find Processes. Ideally, it should only display the list of Servers to which you are actually logged in, and if you are only logged in to one, then it should omit this prompt. If you are logged in to more than one server, then it should default to the last server that you accessed.
The Find can be accessed with Ctrl-F. This prompts you for a string to find, and goes to it. However, there is then no option to Find Next.
The Replace can be accessed with Ctrl-Shift-F (Ctrl-H could not be used as in a browser this brings up your browsing history). The first prompt that comes up is Replace which confused me as I thought this meant the text that you wanted to replace the text with, however, it is actually a prompt for the text you want to find. When you enter some text you then get prompted for the replacement text. It would be clearer if both the Find and Replace text boxes were shown at the same time. You then get options for Yes, No, and Stop, however, there is no Replace All option, which would be useful. To Replace All you need to click Yes until all matches are found and replaced.
Of things that you can do in Architect that you cannot do in PAW Modeler, as far as I can see:There does not appear to be a way to save an existing dimension with another same. You can save a Hierarchy as another name but that only saves the new hierarchy within the dimension.
In the dimension editor you cannot change an element from a numeric to a string, or vice-versa. You have to delete it and re-insert it with the right type.
I created a Drill Process. I found that the ReturnViewHandle statement was not automatically present in the process. It only appears when the process is saved and re-edited, and then it appears outside of the Generated Statements area, which is not the case in Architect. I typically need to put an IF( 1 = 0 ) around the generated statement to disable it and then put in my own statement to return the view that I created on the Prolog. It is hard to do that when the statement does not appear initially.
On the Drill Process I got an error 'Attempting to load command DrillThruAction that is not known in the definitions' which is not an error I have seen before and it is a particularly opaque error message. I think it was caused by the lack of the ReturnViewHandle statement.
If you right click on a Control '}' dimension, you cannot Edit Attributes on that dimension. This is something that we quite often want to do, eg to add an attribute to the }Groups dimension, to associate it with a Legal Entity, etc. You can work around this by running a process to do it, but it is something that you can do in Architect.
When you delete a process, it does not automatically refresh the display to show that the process has gone, which is confusing. If there is too much overhead in deleting the process, it should at least grey it out or show it in red or give some indication that it no longer exists.
On a Process, the Variables on the Data Source Tab are now listed horizontally, instead of vertically as they are in Architect. The vertical presentation is easier to read as there are only three column, Variable Name, Example, and Data Type/Ignore . However, most data sources will have multiple variables, and therefore with the horizontal presentation, that means that you need scroll horizontally to see them. It would be better to revert to the vertical presentation used in Architect.
SummaryOverall the new PAW Modeler is nearly there and I do like it, but as with any new product, there are a few things that could be improved. The only real show stopper is the lack of access to TM1ProcessError log and TextOutput files.
Regards
Paul Simon
------------------------------
Paul Simon
------------------------------
#PlanningAnalyticswithWatson