EAF documentation appears to be wrong
Closed this issue · 7 comments
File: ga/pario/doc/api_eaf.html
(or: http://hpc.pnl.gov/globalarrays/pario/EAFapi.html)
bytes
is always described as double
, while it sure looks like an integer in the code and examples.
In EAF_Read
, there is a "integer req_id" (which is not an argument), and offset
is described as "Absolute offset, in bytes, to start writing at", but I don't believe that (should be reading, not writing).
The test.F
file uses integers.
Are we looking at the same code?
https://github.com/GlobalArrays/ga/blob/master/pario/eaf/test.F#L280:
double precision length, offset, ofw
https://github.com/GlobalArrays/ga/blob/master/pario/eaf/test.F#L312:
ierr = eaf_write(fd, offset, buf, nwrite)
Yes, but bytes
is the 4th argument, i.e., nwrite
, and nwrite
is an integer (line 279):
integer avail, fd, i, nwrite, ierr, iolen4, iolen, loop
I guess you assumed I was talking about offset
, which is indeed a double, no problem with that.
Is https://github.com/GlobalArrays/ga/blob/master/pario/eaf/eaf.doc correct? If so, either the HTML docs can be modified or just deleted. The HTML was created with https://en.wikipedia.org/wiki/Adobe_PageMill, which was active between 1999 and 2000. Moving the docs to GitHub Markdown makes a lot more sense than HTML hacking...
At least it does not have the same mistake, because it doesn't describe the arguments.
Closing due to inactivity, feel free to re-open if needed.