Thanks Cynthia, Would you even recommend against just pre-pending a path to the current autocall search path (like this macro from Peter Crawford)? That is, use soft code to determine the current autocall option in effect from whatever the SAS admin has defined, in my case would return:   ( "SASEnvironment/SASMacro" '!SASROOT/sasautos' ) And then pre-pend my personal autocall library to that:   ("/home/mymacros"  "SASEnvironment/SASMacro" '!SASROOT/sasautos' ) ? In non-BI SAS, I think lots of people are in situations where there are hierarchies of macro libaries (SAS supplied->Corporate Macros->Project Macros), would be a shame in BI to give up on having project-specific autocall libraries.  (And I would feel the same for pre-pending to fmtsearch to look for project-specific formats before looking for any server-level formats defined by an admin). --Q. ... View more

@ckx, if you want to catch notes re uninitialized and missing values generated, one way is with the (undocumented):   options dsoptions=note2err; It turns bad notes into errors.  I love it.  I submitted to SAS ballot to add an item to recommend that they make this a supported option.  Not sure if it made it. But I'm hoping the more people start using it, the more likely SAS will make it supported. I learned about it from Andy Ratcliffe's blog:   http://www.notecolon.info/2009/10/note-more-about-note2err-aka-be-of-good.html ... View more

Thanks Chris, As soon as I returned to working after writing that post, I noticed that Import Data Task.  And was curious as to what magic allowed it to read data from an Excel file sitting on my windows PC and import it into a SAS session running on a Linux server.  And then looked at the log and it said, basically "fancy stuff happened here that isn't reflected in the log."  : ) OT: So now I'm back to thinking that if I want a stored process that will read an access database sitting on a windows file server, I need to set up a windows job to periodically copy the file to the linux server (samba on the server allows windows to see the directories) then read in the file using the PCFILES engine I suppose.... --Q. ... View more

Hi Doc, I'm not sure you can blame the OS for EG not letting you know when a file has been modified by something else.  For example on windows, I can open a .sas file in UltraEdit and EG.  If I edit the file in EG and hit save, the second I hit save UE pops up a dialog box that the file has been revised, asking me if I want to reload (before I've even shifted focus to UE!).  If I edit the file in UE, EG does not reflect the update, and will not reflect it until I close the program and re-open.  If I save the file in EG, the edit I made it UE is lost.  Even MS Word handles it better than EG, since Word tries to get a lock on a file.  Which is not an argument for opening a file in multiple apps at the same time.  I've just got so used to UE's notifications, I've come to expect it.... As I'm new to EG, curious as to what tasks you are referring to that can only be run in EG? My understanding is all the EG Tasks are producing SAS code in the background, so all the SAS code generated by EG could be written by a person, and almost all the SAS code generated by EG could be decoupled from EG.  (I say "almost" as I imagine code generated by EG may rely on some automatic macro vars/environment vars that exist only in EG).  I like the idea of having a program that I can run in EG, or locally on my PC in DMS, or batch on my PC, or batch on the server, or ..... Thanks, --Q. ... View more

Thanks Tom, That's helpful.  Since I will be working in a BI Server environment, sounds like for oracle connections I might just define the full connection info in the libname statement.  I would be creating shared librefs defined in the server metadata.  So even if a DBA moved the database, I would only need to update that one library definition to get the library working for all users.  At least it's good to know that I will (should) be able to create new connections to an oracle database without having to submit a ticket to the help desk.  I assume there will only be one TSNAMES file sitting on the server, and I won't have write access. Sounds like for SQLSERVER I may be stuck submitting a help desk ticket to have them add lines to the odbc.ini file every time I want to connect to a new database. sigh. Thanks again, -Q. ... View more

Hi, Well you've got 10 years more SAS experience than I do (and I've only been using EG for a month or so), but I would say... I don't really see need for batch mode in EG.  That is, if I want batch mode, I can always open a terminal session on the server, and batch submit there.  Or no doubt could even set up UltraEdit to batch submit to the server.  I think EG is an interactive environment, just like DMS is an interactive environment .  (This may be related that I use EG similar to DM SAS, with just one program, or one program that calls other programs via %includes, rather than using the EG process flow to link small code modules. So in the end I have a program I can batch submit without needing help from EG.) Strongly agree with the --EEFILECHANGEUPDATES point.  I've had problems where I had a program opened in EG, edited in Ultra Edit, and expected to see the edits in EG (cuz it's a link, right???) but nope.  Even within EG, I've started opening the actual stored process .sas file in EG.  If I modify the stored process, the modified code won't be reflected in the .sas file I have open unless I close the project (not just the program) and re-open.  This has caused me some heart-ache! Agree about keys for controlling windows.  In another thread ( https://communities.sas.com/message/128833#128833 ) Chris H. pointed me toward using ctrl-tab to cycle windows, but I still miss the ability to define function keys for myself. --Q. ... View more

Thanks Shiva, I guess my question is really: As a SAS programmer, should I be able to create a new connection (libref or whatever) to a new SQL server database / Oracle database without involving a network admin? So for SQL server, looks like the odbc.ini file needs to be revised to create a new connection.  If I have access to SAS Management Console, is there a tool in that will create the necessary updates to the odbc.ini file?  Or do I literally need to hop over to the linux box, and edit the file directly.  If it's the former, it would feel to me like SAS programmers can create connections by themselves (i.e. using only SAS).  If it's the latter, then I would say SAS programmers cannot create new connections by themselves.  Which would be disappointing, as we pull data from new databases all the time. Same question applies to Oracle. Thanks again, --Q. ... View more

Hi, One option is to run Samba on the UNIX server.  I haven't administered it, but as a user the end result is that the autocall library is sitting on the Unix box, but in windows you can map a drive to it. So the macro definitions might sit in /autocall And your PC config.sas would point to Q:/autocall. That worked great (only gotcha is that on unix, the filenames in the autocall library have to be lowcase!). I don't have Samba where I'm set up now, so agree it would be nice to have a solution for this.  Currently I just copy files from Windos to Unix when I update the macro library, which isn't that often.... --Q. ... View more

Hi, Not sure what Matthew means by changing it to a macro. As I understand it, when you execute a stored process from EG (or elsewhere) it runs on a server, and has it's own WORK library.  It does not use the same work library as the EG session.  See for example.: https://communities.sas.com/message/8585#8585 I suppose it's possible that in your EG session you could find the physical path to the work directory: %put %sysfunc(pathname(work)); And then you could pass that as a parameter when you invoke the stored process. But that would feel like a kludge.  Probably better to just put whatever data you want the stored process to see in a permanent directory that the stored process can see. --Q. ... View more

Just eyeballing things, it seems like the task status of "collecting" often takes the longest time, consistent with what Chris wrote.  I assume this is when EG downloads the output and the log for display.  Even just simple html output is a notable lag. (Of course I'm new to EG, and used to running just DMS locally, where there is no lag).  Are there debugging options that would allow us to get more information on these transfer times, like a souped-up version of FULLSTIMER that would include times for uploading the program to the sever, and downloading the results and log from the server? That might be helpful to SAS admins in tuning things.  Just a thought.  Perhaps it's already there in the server logs, I don't have access to those... ... View more