The current data acquisition system and observing environment at CFHT, Pegasus, was a major step forward when it was implemented in the early 1990's. New types of instruments and other new requirements have now made it essential for CFHT to develop a replacement for Pegasus. The primary goal of the new system (code name GenX) is observing efficiency. GenX is will be designed to effectively integrate TCS and instrument control, support queue scheduling, allow for true remote observing, i.e. from Canada or France, and allow for efficient operation of Megacam. This paper will briefly describe some of the motivation for moving to GenX, outline the major goals of the system, describe some of the concepts being considered and some proto-typing which has been done this far at CFHT which may be incorporated into GenX.
When Pegasus was developed we were several years away from many of the software technologies commonly used today: Web-browsers, Tcl/Tk, Java, CORBA are some of the technologies used in most other observatories and by the software industry in general. These technologies make it possible to develop a replacement system which will be faster to develop, provide for a more efficient observing environment and meet the new requirements from more complex instruments and observing modes.
CFHT has made improvements to the Pegasus system to take advantage of new software systems and to meet the requirements of new instruments. For example, the Gecko interface is now based upon Netscape, replacing the Xforms used by Pegasus. The underlying control system still uses Pegasus, but the use of Netscape for the interface has provided a better interface which is also much easier to expand and modify.
In this paper we will describe the scope of the GenX project, the general goals of the project, some of the technologies being considered and the general capabilities of the system. Also, the general philosophical approach we expect to take to GenX development will be briefly discussed.
Only major CFHT instruments will be migrated to use the GenX system. Instruments such as the FTS will be left as they are since their expected lifetime and level of use, do not justify the amount of work needed to include them in the scope of the project.
Detector (both CCD and IR) control is included in GenX and in fact much work has already been done to move detector control away from Pegasus. The DetI system for control of CCDs and IR arrays developed by Jean-Charles Cuillandre for KIR is a more efficient system for both observing and engineering than the detector control paradigm used in Pegasus. This work can be considered a (very effective) proto-type of concepts and approaches which will be used in GenX.
GenX will see much better integration between the data acquisition and telescope control systems. The architecture of the new TCS IV system allows for much better communication and coordination between the data acquisition environment and the TCS. This should allow for more efficient acquisition of targets, guide stars and calibration sequences.
GenX will include support for queue scheduled/service mode observations. These will require the principal investigator (PI) to define the observations for the project in terms of observing blocks or sequences. Thus GenX will be capable of handling command sequences or scripts.
Remote observing will be supported by GenX. This not only includes observations made in Waimea, which are not really remote because of the high bandwidth available, but also from sites in Canada and France.
Efficient archiving of CFHT data and effective integration with the Terapix project for Megacam data is also a requirement of the GenX project.
GenX will not include replacement of the low-level instrument control S/W such as DUCK, although this may need to be modified to interface with other GenX systems. Also, while changes and extensions to TCS IV will have to be made to integrate with GenX, no major development of the TCS is foreseen.
As mentioned in the Section 1 the overriding goal of GenX is increased observing efficiency for traditional observing. Observing efficiency is defined as maximizing the amount of open shutter time on scientific targets. Optimizing observing efficiency for new observing modes such as remote and queue scheduled is also a prime goal of the system. While S/W systems play a major role in the efficiency of any data acquisition system, other aspects are also very important. For example, requiring observers to submit target lists in advance, with CFHT then identifying guide stars in advance of the run, will also increase observing efficiency.
It is possible to measure CFHT observing efficiency as all exposure information is included in the archive database at the CADC/HIA in Victoria. It will be possible to quantitatively evaluate GenX against its primary goal.
What can appear as fairly small gains in efficiency may actually be significant gains. Saving 30 seconds between exposures translates to approximately 30 minutes saved over the night. Thirty minutes saved each clear night at CFHT translates into the equivalent of more than 10 additional nights of observations. While saving 30 minutes each night does not really result in extra nights in the classically scheduled mode, in queue scheduled mode this extra time is real in the sense that more programs can be completed.
GenX is being designed to increase observing efficiency and new capabilities beyond those of Pegasus to this end. Introducing new technologies will be need to realize many of these new capabilities.
Consider the goal of efficiently supporting remote, i.e. off island, observing. The Netscape browser interface for Pegasus provides this capability with very little S/W development. While Netscape, or another Web browser, may be limited as a user interface it offers several advantages. It runs on several different platforms, it is familiar to most, if not all, of our potential users, it needs very little if any S/W support and it is unlikely to disappear any time in the near future. If an instrument requires more capability than to work efficiently with a browser then Java applets can be used to extend the browser capabilities.
GUIs such as Netscape are easier for users who may not use an instrument very often or for more complex instruments. However, for experienced users, a command line interface can actually be more efficient. GenX will have both command line and graphical interfaces (CLI and GUI) which can be used simultaneously, as with KIR and CFH12k. This allows users to select the interface they are most comfortable with or to even move between the interfaces if, for instance, they forget a command name. Command line interfaces also allow for easy scripting of commands to allow for automated sequences of commands. CFHT's new DetI/Director interface (Director is the creation of Sidik Isani at CFHT) for KIR and the CFH12k interface provide this capability.
While operating (i.e. sending commands and receiving status information) an instrument remotely is fairly straightforward, dealing with the data in an effective manner is still a challenge with the bandwidths generally available between CFHT and sites in Canada and France. We have some ideas on how to handle the display of data to allow for efficient remote observation. Generally, it is very rare that an observer needs to examine a complete image with full fidelity and resolution. Often, one simply needs to examine a relatively small subraster or to look at the complete image in a general sense. We believe that an intelligent display system, consisting of a display server and display client, with the client running at the remote end, can effectively support remote observation.
This display system would incorporate compression techniques, both
lossy and non-lossy, between the server and client. For example,
when an image was initially displayed a lossy compression algorithm,
such as ``hcompress'' used on the Digital Sky Survey, would be used to
compress the data being sent from the server. The client would uncompress
the data when it was received and before being displayed. Compression
factors of 100 would still allow for field identification, OSIS guide
star selection, etc. An 8 MB image from a 
CCD would be
reduced to 80 KB. The compression would occur on-the-fly while
the CCD was being read, this introducing minimal overhead. The system would
also be able to handle subrasters intelligently. The remote user would be able
to select a subraster whose definition is sent to the server which could
then send that subraster, uncompressed, for a detailed examination by the
user. A 
subraster represents
40 KB of data to send over the network.
Of course, this system would be unable to support all instruments. It
will be several years before one could think of operating Megacam from
France without some on-site support.
GenX will provide the user with continually updated status information on the detector, instrument, telescope, weather conditions, etc. A single, comprehensive and configurable status display will allow the astronomer and the Observing Assistant to determine the state of the observing environment quickly.
GenX will allow the observer, including remote ones, to move the telescope between targets when it can be done safely. This would include moving the guider to acquire guide star which may be possible to do automatically. This would allow for more efficient observing for remote observers and make it easier for one person to operate both the telescope and instrument.
GenX will work efficiently off previously entered target lists which can include guide stars. As the telescope moves to the new target the guider, offset, OSIS probe, AOB WFS, will be moving to position themselves close to the correct position. The final accuracy in this positioning will be determined by the TCS system and will be on the order of 10 arcseconds.
GenX will provide an integrated quick-look processing environment based upon IRAF. IRAF provides most, if not all, of the basic functionality needed and can be extended as necessary with scripts. Such as quick-look processing environment would also be part of the remote observing environment with IRAF running in Hawaii so the data never have to be sent over the network.
A very short note about the development philosophy we will be pursuing. CFHT intends to re-use parts of Pegasus and public domain S/W or S/W which may be available from other observatories if we feel it is useful and makes sense.
We also intend to keep the complexity of the system to the minimum which is required to get the job done! This point may often be in conflict with the first point raised in this section. Is it better to reuse available S/W which may be more complex and provide many more capabilities than needed, or to develop the software ourselves?
GenX will be the new data acquisition S/W for CFHT which will provide an efficient observing environment for the current modes of operation as well as future modes such as remote observing and queue scheduled observations. The primary goal of GenX is to maximize the efficiency of CFHT as measured by open shutter time from twilight to twilight!
