0.9 fix ODBC and DBI interfaces for bigglm to not use LIMIT, and just not allow
    variables to be floating free in the workspace (which really couldn't work
    anyway)

    fix arguably-false-positive from Fortran bounds checking, by incorporating
    the fix in the published AS274

    S3 methods added to namespace, and Fortran syntax updated to be 2018-compatible

0.8 allow offsets in model formulas for both biglm and bigglm.

0.7 Proper S4 inheritance for DBIConnection method. There's still something
    wrong with the set-up, since we have to export all the S3 methods for it
    to work.
  
    summary.biglm object now has $nullrss, $rsq components



0.6 bigglm() now warns clearly when it runs out of iterations
    before convergence. (problem reported by Francisco J. Zagmutt)

0.5 SQLiteConnection method (RSQLite package) for bigglm.

    biglm() no longer gives an error message when used on 
    data of storage mode 'integer'

    [The leaps package now has a regsubsets() method for biglm
    objects (for Tal Galili).]

    Sandwich variance estimates were wrong when weights were
    specified. 

    predict() method and extractors for AIC, deviance, based
    on code from Christophe Dutang

    RODBC method (RODBC package) for bigglm

0.4 bigglm for glms.

0.3 Added sandwich estimator

0.2 Added weights

0.1 Initial version