See ~gfisher/projects/work/support/documentation/LOG-template for a description of the fields of the LOG file table and conventions for making LOG file entries. DATE AUTHOR STATUS FILE(S) DESCRIPTION ============================================================================== 5aug99d GLF TODO convetions/? Consider adding rationnale for various conventions, either in the convention docs themselves or in (a) separate rationale doc(s). 5aug99c GLF TODO conventions/ Per the last paragraph in the design-imple description of 5aug99a item below, update the conventions for the non-use of pure virtual functions. Although the conventions generally do not include much rationnale, we could add some in this case. 5aug99b GLF TODO LOG-template Change, once again, the title of the and all LOG "DO" status item to "DO IT". The files every- latter really does seem to read better where! (but haven't we been here before?). Anyway, I really do think that "DO IT" is better, and will hopefully continue to think this, at least for a while. 5aug99a GLF TODO list.{h,C} Changed pure virtual defs of four DONE functions to noop implementations. The today functions are: ListElem::Compare(ListElem*), ListElem::Compare(ListElemKey*), ListElem::GetKey(), and ListElem::Print(). The motivation was the compilation of the revised rbrowse code, which was updated to use the support lib List instead of the locally defined list it had been using until now. With the pure virtual function defs, the expected pain-in-the you-know-what problem happened during compilation, since rbrowse had never had to specialize any of the functions that were pure virtual in the library, since the local list def did not use pure virtuals. MORAL OF THE STORY: The convention never to use pure virtuals is an excellent convention that should always be followed. The wording used in the comment of the previously pure virtual functions can be part of the conventions. Viz., it says the following for any would-be pure virtual fucntion F: "F has a noop default implementation, in particular it uncondtionally returns NULL. F must be specialized to produce meaningful output." Note that the "unconditionally returns NULL" clause is omitted for void functions. 14mar99 GLF TODO admin/{copy, Ignore SUN4*-work-work as well as update}-work SUN4*-work. -to-alpha 28feb99 GLF TODO config,admin Make a new config dir and move current config-related admin files there. High on the list of such files are the ones dealing with making. 16feb99 GLF TODO conventions/ Add convention about using "class XXX" design-imple vs "#include xxx.h". Viz., use the former whenever it will work. 9feb99 GLF BUG sample-Makefile Changed makedepend path to normal FIXED system path (i.e., no explicit path) today from the version in ~gfisher/bin. Last year sometime the need arose to use the version in ~/bin, but the problem seems not to occur anymore. Details of the problem are in the comments in ~/bin/makdepend. 10aug98 GLF TODO idraw/* Add a SelectVisibleObjects, bound to, say, ctrl-left, that selects only the fully visible objects on the screen. The modivation is to be able to grab previously grouped scenes that were ungrouped on the top of other ungrouped objects. The long-desired regroup op may make this form of select superflouous, but I think I'd like to have both. 29jul98 GLF TODO conventions/ Update var naming conventions as design-imple follows: * No more short names * Truncate long names where possible * Make it all make sense somehow * Give plenty of examples 2mar98 GLF TODO list.{h,C} Add a FindEnum(ListElemKey*) and a ResetFindEnum() that will enumerate all found of the given key. GLF TODO list.h Expand the documentation to explain std specialization better, in particular the specialization of keys and how sort will work for free if Compare is properly implemented. GLF TODO doc/ Add conventions/ sys-specs.me