    CHANGES/IMPROVEMENT MADE TO NCLEVER

Version 4.01
    February 1999
      - Fixed the behavior of the "saveuid" command when the current
        list of document is empty; if option AllowNull is TRUE, it
        will save an empty list.

Version 4.0
    February 1999
      - Rewrote the main search engine to use the EntrezTlEvalStringX()
        function. All query terms are now maintained as text strings
        such as '"rioux" [AUTH]' and '"glucagon" [WORD]'. Boolean
        operations are performed by building longer strings which contains
        those terms linked by the operators "&", "|" and "-". Therefore,
        the sequence of nclever commands

            AUTHOR rioux
            TEXT portable

        will result in the construction of this query string:

            "rioux" [AUTH] & "portable" [WORD]

      - (Technical internal stuff:) Fields are no longer specified with
        FLD_xxx codes, instead [XXX] tags are used; GetEntrezInfo() used
        for field validation.

      - Added three new date search commands: "EDATE" (Entrez Date),
        "MDATE" (Modification Date) and "PDATE" (Publication Date).
        The normal DATE command is synonymous to EDATE.

      - Added a special "SEARCH" command which allows the user to explicitely
        enter a text query string as a single term, even one with boolean
        operators. So it is possible to do in a single stroke:

            SEARCH "rioux" [AUTH] & "portable" [WORD]

        This also allows the user to search in indexes that are not
        currently supported with the set of nclever search commands,
        such as '"123-135" [PAGE]' to search by page number.

      - Added the "INFO" command, which lists the currently supported
        indexes at NCBI, for each database. The text tags are also
        shown, which is useful when using the new "SEARCH" command.

      - Added support the Entrez Browser's new format for saved uid lists,
        where the first line which identifies the target database is one
        of ">Medline", ">Protein" or ">Nucleotide" instead of the
        numbers 0, 1 or 2. Nclever can still read the old format, but will
        always write in the new one.

      - Everything related to the "Special" count in entrez indexes has
        been removed; for compatibility reasons the program will
        silently ignore the old "FetchSpecial" option when supplied
        interactively, in the .cleverrc configuration file, or with
        the -s switch after a search command. No harm done.

      - For protein sequence documents, the "GenPept" output format is
        now supported.

      - For sequence documents, the "report" output format is no longer
        supported. When chosen with the "report" command, this will have
        the same effect as choosing "genbank" or "genpept".

      - As a consequence of no longer having a "report" format, the
        "ShowSequence" boolean option has been made obsolete. It
        is now completely ignored, like the FetchSpecial option.

      - Truncation mode is implemented as it should, but it DOESN'T WORK
        because of restrictions on the NCBI side. They say it should come
        back later this year. In the meantime, don't use it, or else
        you'll get ugly messages on STDERR. The truncation mode option
        now defaults to "FALSE".

      - There is no support for the CD-ROM version of Entrez. I guess it
        would still work, but I changed all references of "clever" to
        "nclever" in the documentation and removed everything that talked
        about the CD-ROM version. However, the user's configuration file
        is still called ".cleverrc".

      - The distribution no longer has a nroff-formatted man page.

      - The DATA and ASNLOAD external directories whose location
        used to be specified in a .ncbirc file are no longer necessary.

      - Fixed a Y2K bug that made nclever reboot the entire internet
        network across all of north america and europe. Just kidding.

Version 3.1 (still still)
    December 1996
      - Added the possibility of supplying multiple numeric arguments to the
        "taxonomy down" command, as in

        NCLEVER> taxonomy down 3 3 5 12 1 2 5 2 14 1 6

        when you know exactly where you want to go.

Version 3.1 (still)
    January 1996
      - Added the AllowNull option
      - Fixed a bug in the SaveUIDS command (would not detect write errors).

Version 3.1
    September 20, 1995
      - Recompiled using the lastest NCBI toolkit.
      - Version number no longer follows NCBI releases... :-|
      - Changed the GenBank output routine in print.c, since
        the NCBI toolkit has added a more elegant way of doing it.
        BUG: The "showseq" option seems to be ignored for genbank output
        format (sequences are always shown).

Version 3.03
    February 3, 1995
      - Added the "Date" command, since Entrez now support this search field.

Version 3.02
    January 19, 1995
      - Worked around limitation in NCBI toolkit causing segmentation faults.

    December 20, 1994
      - Added the "all" keyword to the "pick", "not" and "union" commands.

Version 3.01
      - This version is the same as version 2.131 and was renamed 3.01 for the
        sole purpose of keeping the version numbers in sync with the
        NCBI browser.

Version 2.131
    December 7, 1994
      - Recompiled using the 7 December 1994 NCBI toolkit (NCBI announced bug
        fixes).

Version 2.13
    November 7, 1994
      - Recompiled using the lastest NCBI toolkit.
      - Patched a strange bug: the EntrezDocSumListGet sometimes passes
        a NULL pointer to the callback routine, when it's not supposed to.
        This now results in printing "DOCUMENT MISSING" rather than segmentation
        faulting. This problem was related to the entrez database.

Version 2.00
    May 1994,
      - Added taxonomic browsing commands.
      - Added the NOT command.

Version 1.07 
    March 31, 1994
      - Changed the prompt from "CLEVER>" to "NCLEVER>" for the "network"
        version.

    February 2, 1994
      - Fixed a bug: saving multiple sequences in ASN.1 format would
        result in only the first sequence being saved. Thus,
        "save my_file 3 4 5" would only save sequence 3.

Version 1.06
      - First released version.
