First off, I'd like to say that I'm a big fan of EG, it's a great front end for getting the bulk of your tasks set up really quickly.
Lately though, I've started to find a number of things that are really starting to get on my nerves a bit. Things which don't quite seem to work right, or could be better.
The environment I'm working in here is more of a client based operation, I have many clients and many different jobs all using different data sent through in various formats. As a result, each project is in a different directory and the libraries and files are all obviously different too. Whereas working on a Data Warehouse with SAS/EG is a joy, this environment sometimes makes EG feel a bit clunky.
Here's a list I've been compiling over the last week. I'm posting this here because I'm not really sure where this sort of thing should go:-
- Clicking on multiple icons (eg selecting nodes one by one holding down ctrl) gets mistakenly identified as a double click on the second icon if you go too quick.
- Occasionally, after a failure, nodes suddenly get linked to pieces of data that they have no connection to. Clearing them can be difficult and usually requires rerunning the nodes and then refreshing.
- When exporting a dataset as a text file as a node in a project, you can only call a TAB delimited file .TAB, not .TXT. If you then rename the file outside of EG, the project cannot be saved without generating an error.
- Need a way to tell SAS not to include output in the project for when outputting large data files in code using ODS (ie ODS CSV) as otherwise it keeps generating a warning message, every time you open the project or refresh the process flow.
- Occasionally on completion of a large piece of code, all the output is missing from the process flow, refreshing doesn't return it.
- Sometimes when in a code window, hitting F3 submits the entire process flow, meaning you lose all your logs etc.
- Stopping a SQL step doesn't release the lock on the files. Have to close EG and restart.
- Stopping is just unreliable, it frequently messes up the process flow, resulting in nodes being joined to incorrect data
- When you import a text document, if you use a column name which is different from the column name picked up from the text file, it doesn't retain this change between runs of the node.
- When you import a text document, if you change the default formats etc, it sometimes doesn't remember these if you re-open the node. Very annoying if you have changed a large number of fields.
- Adding your own colour scheme or output template is a fantastic feature. Shame that it then puts a warning message in EVERY log, even if it isn't actually producing any output, meaning you have to go through you logs every time to make sure that is the only warning.
- Would be nice to be able to set up libraries for a project that get assigned when the project is opened, also set up a default directory, system options, macro variables, autoexec code etc. This would be useful for those of us who do lots of smaller jobs using different data sources.
- Would be nice to be able to STOP ALL running tasks for situations when the entire process flow gets submitted by mistake.
- Would be nice to be able to search all code in a project, rather than just the currently opened node. This feature in VB has always been fantastic.
- Would be nice to have better integration with Macros, a "Task List"-like menu allowing users to select macros to run would be great, especially for those of us who don't have a SAS metadata server so can't use Stored Processes.
- Would be nice to be able to specify default formats and informats for text importing. For instance, the default for a recognised Date field is MMDDYY10., however outside of the US, DDMMYY10. is used almost exclusively. Importing a file with many Date fields can be frustrating, especially coupled with the glitch which loses the specified formats when you reopen the import step.
Thanks for listening.
Message was edited by: ChrisW at Mar 30, 2007 2:44 AM
Added the template annoyance bit as I just remembered it :-)
PS - I know that some of the things under "Glitches" are inexact, and not much help to you in regards to re-creating them, they are just things that occasionally happen to me with nothing particular that seems to have caused it, and not that I could reproduce regularly, they just seem like random things. I'm also aware that some of these are actually listed as known issues.
I have a large list of things I'd like to change, mainly from the perspective of using EG to perform analysis of real-world data so my list would have little overlap with yours.
However, one caught my eye:
- Would be nice to be able to specify default formats and informats for text importing. For instance, the default for a recognised Date field is MMDDYY10., however outside of the US, DDMMYY10. is used almost exclusively.
Would be even nicer if EG was smart enough to tell an MMDDYY from a DDMMYY.
we have an option for the SAS9 anydtXXX informats, clarifying the preferred layout when the numbers leave uncertainty ~ that might be relevant to Proc Import.
However, I would like it to be easier to swap to using datastep code instead of adding options to proc import. The documentation for Data Step is as "enriching", as the capabilities of the code. So, it becomes easier to deal with the complexity of a more sophisticated data step, than the options needed to control the "wizard" import.