This commit removes the #ifdef _DEBUG statements at the top of all source code files per issue #482. It also updates the doc files to stress that the speedup observed for hydraulic analysis with the MMD node re-ordering method only applies to single period runs.
In addition to addressing compiler warnings, argument names in the API function prototypes were made more consistent and descriptive. Also additional Doxygen comments were added in preparation for producing a more thorough documentation of the API.
1. Added a standard header to each code module and removed obsolete comments.
2. Re-named several of the sub-structs in the project struct and re-arranged some of their contents.
3. Re-named _defaultModel to _defaultProject.
4. Removed the need to call EN_createproject and EN_deleteproject when working with the default project.
5. Made X & Y coords. part of Snode properties instead of a separate struct.
6. Moved the non-API functions in epanet.c into a new module named project.c.
7. Re-factored the quality module so that it uses the same nodal adjacency lists as the hydraulics solver.
8. Re-factored the sparse matrix module (smatrix.c) to be more memory efficient.
9. Restricted line lengths to < 90 columns.
10. Grouped the placement of functions in EPANET2.H and EPANET.C by category.
- Deleting controls with node/link deletion made conditional.
- New EN_addrule function added along with a test file.
- Rule structures re-named & rules.c heavily modified.
- Issue with exceeding limit on number of temporary file names fixed.
- VB declaration and DEF files updated.
-Split off calc. of a pump's curve coeffs. from getpumpparams to updatepumpparams.
- Fixed problem with initializing node demand category when default demand pattern set before node is created.
- Increased MAXLINE to 1024 to accomodate increase in MAXMSG.
- Replaced EN_geterror calls in input2.c with geterrmsg (it's bad form to call API functions in the body of the library)
A bug in the delete function in hash.c was preventing the unit test of the new set ID API functions from running properly. In fixing this bug the entire hash table code was refactored to make it look more like the mempool service routines. Also the need to copy the string passed into the table's insert function was eliminated.
Adds new options HEADERROR and FLOWCHANGE to provide more rigorous criteria for hydraulic convergence. Also breaks HYDRAUL.C into 3 separate files to improve code readability.
Lew Rossman