From lfini@chronos Wed Mar 31 11:02:36 1993 Status: RO X-VM-v5-Data: ([nil nil nil nil nil nil nil nil nil] ["6924" "Tue" "30" "March" "1993" "13:24:17" "GMT" "Luca Fini" "lfini@chronos " nil "146" "SUMMARY: Opinions on standards" "^From:" nil nil "3"]) Newsgroups: sci.astro Reply-To: lfini@arcetri.astro.it Organization: Osservatorio di Arcetri X-Newsreader: TIN [version 1.1 PL6] From: lfini@chronos (Luca Fini) Subject: SUMMARY: Opinions on standards Date: Tue, 30 Mar 1993 13:24:17 GMT Thanks to those who responded to my original posting: Eric W. Greisen, Wayne H. Warren Jr., Donald C. Wells Here is my summary of (and inevitably opinion on) the discussion. I hope that the original authors will not fell too much betrayed by my interpolation of their arguments. The first point is that the enforcing of standards is a way to rationalize the work and to make life simpler: >> (...) No telescope in the world >> stands alone any longer so there is no reason for its users to >> tolerate getting data in formats that are unreadable except to >> special-purpose programs. And this should be readable in considerable >> detail. (...) But, moreover, it is a *necessary* approach for doing science: >> but I can't do science if I do not know the coordinates, brightness >> scales, etc. (...) >> "The purpose of standardization is not to enforce the common >> mediocrity, but rather to aid the creative craftsman." as it has been in the past: >> .... most common observations, such as photometry, spectroscopy, >> radial and rotational velocities, etc., employ the use of standard >> stars to match observational systems. .... One of the comments goes a little further on the long debated issue about calibrated data: >> (...) Another standard that should be pushed much more than it >> is is the removal of the "instrumental signature" from the data. >> I.e., the flux and coordinate scales should be corrected (regridded >> and non-linearly scaled as needed) so that the output images are in >> rational defined units and image projections. I understand that this >> cannot always be done in detail - especially say deconvolving images >> to remove synthesis beam patterns or the patterns produced by seconday >> support struts etc - but some efforts should be made in all cases. >> And a fully corrected (if no longer quite exactly right) image should >> be available to the user. Now dealing with specific examples of standards in astronomy, not surprisingly, FITS comes out as the most wide spread and appreciated one, with Tex and LaTeX coming seconds: >> As one of the original authors of FITS and currently developing >> standards for coordinates in FITS I am a great believer in enforced >> standards for data intercommunication. (...) >> (...) FITS format for the inter- >> change of data is a good example of widespread use of a standard, as >> is TeX and LaTeX for document processing. (...) >> I am sure that it will come as no surprise to you to hear that I >> strongly favor astronomy's data format standard, FITS! I am also >> strongly in favor of certain computing standards, e.g. IEEE-FP, >> Postscript, X-windows, TCP/IP, TeX, etc. (...) Then comes the stressing of the importance of standard programming techniques in software development: >> (...) Software too is now a >> community resource - it costs a lot to produce and should be shared >> whenever possible. Rational and acreful standards of conduct in >> coding are needed to make that work. I and others have to figure out >> my code of today in another N years or in another place. >> I believe too many astronomers (...) >> (...) miss the point about the need for standards to enhance >> the spatial and temporal longevity and value of their work. >> (...) Standard languages are also >> important, of course. A comment about importance and ways to enforce standards: >> (...) It is important that astronomers use a >> consistent, documented, interoperable *canonical* data format for >> initial data distribution as well as for longterm archival storage. In >> particular, the space agencies of the world should *enforce* the use >> of FITS for space mission data products. The major groundbased >> observatories should do likewise, especially for large telescope >> construction projects. (...) Now coming to problems concerning the difference between de-facto and de-iure: [referring to data analysis, data archiving, etc.] >> (...) Most of the items that you >> list above, although having no "formal" standards, are represented by >> certain packages and procedures that are widespread and that have >> become de facto standards among astronomers. >> In networking we have a strange situation: ISO's "OSI" protocol suite >> may be the de jure standard, but the de facto standard is the TCP/IP >> protocol suite; one can easily argue that the latter does exist but >> that the former does not! (...) Which suggests how standard should develop: >> (...) OSI failed because it developed >> top-down, out of an intensely political committee process which >> produced cubic meters of paper, but no (interoperable) code. TCP/IP >> succeeded because it developed bottom-up, out of an R&D process which >> produced working implementations and learned from them. The TCP/IP >> process was an example of what is today often called "rapid >> prototyping". The Internet Activities Board (IAB), which is the body >> which holds formal authority over TCP/IP, will not endorse any >> experimental protocol as a TCP/IP standard until interoperable >> implementations have been demonstrated. (...) I would add to this that when a de-facto standard proves to be a good one, it will easily develop into a more formalized one (de-iure). Less likely vice-versa. Finally some remarks on areas where a stronger standardization effort would prove fruitful: >> I think that the computing industry needs to invent an external >> canonical format to interchange tables between database systems. Each >> system has its own external format. The FITS BINTABLE extension is a >> sufficient solution for astronomical computing people. >> [remote archive access is an area where] We need to have a small set >> of standards. I am very impressed with the effectiveness of WAIS >> (Z39.50): it is far from perfect, but the fact that already such a >> diverse set of archives can be searched with a single portable client >> user tool is an impressive demonstration of the synergistic power of >> standards -- such standards create *markets* which encourage growing >> sets of entrepreneurs to provide new services for growing sets of >> customers. +------------------------------------------------------------------------+ | Luca Fini | | Osservatorio Astrofisico di Arcetri Tel. +55 2752 253 | | L.go E.Fermi, 5 Fax.: +55 220039 | | 50125 Firenze e-mail: lfini@arcetri.astro.it | | Italia | +------------------------------------------------------------------------+