Minutes of Software Tools meeting Mar. 8, 1996
            ===============================================

Present:    J. Hobbs, A. Jonckheere, S. Krzywdzinski, L. Lueking, L. Paterno,
            Q. Li, H. Prosper (minute taker) and S. Protopopescu.


    Laura Paterno presented the first draft of recommendations for the new
code development environment. 

    A) General comments/points of agreement

    1. There was general agreement that  the scope of the recommendations should
       be broadened to  include FORTRAN,  to reflect the  recommendations of the
       Languages working group.

    2. The  section on  editor   recommendations  should  defer to  the document
       containg the editor recommendations.


    B) Specific comments/points of agreement

    1. Code  developers should be  provided with  tool to allow  for transparent
       remote compilation.

    2. Some level (as yet unspecified) of  reverse engineering would be helpful.
       Qizhong mentioned the impressive capabilities of the Flint product, which
       however is not  widely available  due to cost. She  mentioned that it was
       available on Fnalo and that she has used it to good effect on D0RECO.

    3. A list of compliant compilers should be provided to potential code
       developers.

    4. A contact person should be assigned to each compliant compiler.

    5. There should be a  list, that is periodically  updated, defining standard
       software services that provide some (as yet unspecified) level of machine
       independence.   An   example of  such  a  generic  service  is  D0open. A
       preprocessor  should be  able to  handle  D0-specific  macros, whose aim,
       again,  is to  provide  machine   independence. The  feeling  was that if
       possible machine blocks should be avoided.

    6. A tool to automate  document extraction from  source code, and to provide
       some level of   "documentation  enforcement" should be  provided.  But, a
       certain healthy  skepticism was  expressed, notably  by John H, about the
       feasibility of "documentation enforcement".

    7. It was  mentioned that Herb  Greenlee  suggested  replacing D0FLAVOR with
       CPP. We agreed that this seemed  sensible. (Editorial comment: We need to
       better understand the  implications of this.  For example, does CPP allow
       us to invent our own macros, or are we forced to use only the C macros?) 


    Alan Jonckeere then  resumed our discussion of  code management. A key point
was  raised:  the need  to  replace all  D0  scripts  (RCP,  RELEASE  etc.) by a
universal    scripting  language.  Alan,  and  others,  noted  that our  current
home-grown   scripting   languages are  very   simple---yet  powerful,  and that
simplicity should be an important aspect of the universal scripting language. It
was agreed  that we  would invite  Scott  Synder to make  a  presentation to our
group, at the Mar 22 meeting, about modern scripting languages.

    It was  agreed  that we  needed to  identify  and list  long-term  jobs, for
example maintaining  the D0-version of  emacs, so that we  have a better idea of
manpower needs.

    There  was the   sentiment that  a  crucial,  unresolved,  issue  was how to
re-organize   D0library. Serban  had  previously  mentioned the  notion of task.
Serban  suggested,  and we  agreed, that for  homework   everyone  should try to
analyze D0library into tasks.