COIN backend should support basis status and tableau data functions

[mkoeppe]

This is for COIN backend support in #18733 and #18688.

Should add
(1) getBasisStatus, setBasisStatus,
(2) getBInvARow, getBInvACol, getBasics,
to src/sage/numerical/backends/coin_backend.{pxd,pyx}

Possible that one needs to add other functions as well, such as enableSimplexInterface...

Reference to the relevant Clp header file:
https://projects.coin-or.org/Osi/browser/trunk/Osi/src/OsiClp/OsiClpSolverInterface.hpp?rev=1472

CC: @yuan-zhou @nathanncohen

Component: numerical

Author: Yuan Zhou

Branch/Commit: c2488a8

Reviewer: Vincent Delecroix

Issue created by migration from https://trac.sagemath.org/ticket/18763

[mkoeppe]

Description changed:

--- 
+++ 
@@ -1,7 +1,12 @@
-Should add `getBInvARow`, `getBInvACol`, `getBasics`
+This is for COIN backend support in #18733 and #18688.
+
+Should add 
+(1) `getBasisStatus`, `setBasisStatus`, 
+(2) `getBInvARow`, `getBInvACol`, `getBasics`,
 to src/sage/numerical/backends/coin_backend.{pxd,pyx}
+
+Possible that one needs to add other functions as well, such as `enableSimplexInterface`...
 
 Reference to the relevant Clp header file:
 https://projects.coin-or.org/Osi/browser/trunk/Osi/src/OsiClp/OsiClpSolverInterface.hpp?rev=1472
-This is for #18733 and #18688.
 

[yuan-zhou]

Branch: u/yzh/coin_backend_should_support_basis_status_and_tableau_data_functions

[sagetrac-git]

Commit: 9fdbd84

[sagetrac-git]

Branch pushed to git repo; I updated commit sha1. New commits:

9fdbd84

Fix bugs in CoinBackend::get/set_basis_status(); Add doctests

[sagetrac-git]

Changed commit from 9fdbd84 to 3644950

[sagetrac-git]

Branch pushed to git repo; I updated commit sha1. New commits:

3644950

Add CoinBackend::get_binva_row, get_binva_col

[videlec]

comment:6

Please fill the Authors field with your full name.

Instead of sage_malloc you could use check_malloc (that checks that the memory is indeed allocated and raises an error if not).

Otherwise, everything looks clean and tests run smoothly on my computer (with Coin installed).

Vincent

[videlec]

Reviewer: Vincent Delecroix

[yuan-zhou]

comment:7

Instead of sage_malloc you could use check_malloc (that checks that the memory is indeed allocated and raises an error if not).

Thanks a lot for reviewing the ticket.
Does one call sage_free to free the memory allocated by check_malloc?
Since sage_malloc is used throughout the file, I was trying to be consistent. Would you suggest replacing them all by check_malloc? Or perhaps better to update them using #18868?

[yuan-zhou]

Author: Yuan Zhou

[videlec]

comment:9

Hi Yuan,

Replying to @yuan-zhou:

Instead of sage_malloc you could use check_malloc (that checks that the memory is indeed allocated and raises an error if not).

Thanks a lot for reviewing the ticket.
Does one call sage_free to free the memory allocated by check_malloc?
Since sage_malloc is used throughout the file, I was trying to be consistent. Would you suggest replacing them all by check_malloc? Or perhaps better to update them using #18868?

I do not think that #18868 is good idea in this case. You will create a Python object for nothing.

check_malloc is equivalent to sage_malloc except that it will raise an error in case the memory was not allocated (where sage_malloc just returns NULL). So the memory should be freed with sage_free. You can safely replace occurrences of sage_malloc by check_malloc. Moreover you can always replace

my_ptr = sage_malloc(12)
if my_ptr == NULL:
    raise MemoryError("allocation failed")

by

my_ptr = check_malloc(12)

Vincent

[sagetrac-git]

Branch pushed to git repo; I updated commit sha1. New commits:

3ab6b33

replace sage_malloc by check_malloc