Long ints, strings and calcout

[pheest]

This PR incorporates and supercedes #30

In dbfield.c, the previous version incorrectly defines INT64 and UINT64 epicsInt32 and epicsUInt32.
This meant that values outside the 32-bit range were lost.
For long strings, the size is not fixed as MAX_STRING_SIZE (40) but variable.
For short strings, addr.field_size is MAX_STRING_SIZE, so remains the same.
For long strings, special runtime processing is required as identified by the SPC_MOD flag.

In devsupapp\src\makefile, the new _dbapi.dbd, _int64.dbd and _lsilso.dbd files are copied to the PY install directory.

In Init.py, these files are used in replacement of the previous use of a temporary database file created on-the-fly.
I have made this change for three reasons:

1. The text is constant (determinstic) so there is no need to create it on-the-fly.
2. There is a need for if clauses on the EPICS base version.
3. The previous use of an on-the-fly file is problematic for Windows file-handling due to the use of simultaneous read and write access.

In test_dnb.py I have added suitable tests for long strings, ints and calcout.

[tynanford]

../dbfield.c: In function ‘pyField_putval’:
../dbfield.c:332:17: warning: suggest parentheses around assignment used as truth value [-Wparentheses]
  332 |             if (prset = dbGetRset(&self->addr))
      |                 ^~~~~

there is a compiler warning here. Could you add the extra parentheses to get rid of it?

[tynanford]

Nice! This looks good to me. Maybe could add a calcout test to test_db.py

I suggest we get the windows PR merged first #51 and then rebase this PR and merge it after? Since both PRs touch devsupApp/src/devsup/__init__.py and create a _dbapi.dbd file.

[pheest]

Maybe could add a calcout test to test_db.py
I thought I had done so. Checking, I see that I only committed it to my 'devel' branch.
I have added the test here also.

then rebase this PR and merge it after
In principle, these PRs could be processed in any order, your choice.
it could make sense to rebase between them, but this does require fairly close co-ordination between us.

[tynanford]

I think Py_DECREF(aval); is required to avoid a memory leak. That was added here: eafb208

[tynanford]

I thought I had done so. Checking, I see that I only committed it to my 'devel' branch.
I have added the test here also.

Thanks!

In principle, these PRs could be processed in any order, your choice.
it could make sense to rebase between them, but this does require fairly close co-ordination between us.

Yes I agree. Well let's plan to merge windows PR first, maybe give it a couple more days in case anyone else wants to weigh in.

It looks like the commit to add the calcout tests made serveral more changes in dbfield.c . Was that intentional?

4b5fbeb

[zhangt58]

It looks like the commit to add the calcout tests made serveral more changes in dbfield.c . Was that intentional?

4b5fbeb

It seems to me that the changes applied to dbfiled.c, most are not necessary, I'd suggest to work with PR #43 merged.