Given that I have been retired from Intel for more than three years, it made sense for me to put the Doctor Fortran blog on my own site rather than on Intel's. So, without further ado, the new location for Doctor Fortran is:
There you can read my latest post along with previous post-retirement posts. I hope to be able to copy pre-retirement posts there as well, but there are links to all. You can also subscribe to be notified by email when a new post appears. (I hope to add one last post on IDZ but am blocked from doing that right now - I expect that will get resolved soon.)
I have read most of the Dr Fortran posts in the past and found them both helpful and informative. It is great to have such good information/resources still available. Thanks.
With Intel's OK, I have now copied all the Doctor Fortran posts I could find, including those from the DVF Newsletter, to the new site. Three posts are gone, unfortunately.
I've also figured out how to show only an excerpt of each post on the home page rather than the entire post, and have opened up comments on all the posts. Please do let me know if there's any topics you'd like the Doctor to address.
Very nice and very useful. THANK YOU again so mush for your service to Fortran community. I am still struggling but, I would not be able to get this far without your help.
See you around
PS: Very sorry to hear Stan passing.
Great article. Is there some way you can influence Intel to include a link to this article in their Fortran documentation?
Question on polymorphic/class types. Are the hidden descriptors defined by a Fortran standard? IOW can one be assured that these descriptors are comparable amongst all compiler vendors? i.e. generating a library from one vendor can be used by a user using different compiler vendor.
The descriptors are NOT standardized and it has long been the case that you can't mix Fortran objects compiled by different compilers. I recall an inquiry, many years back, by one of the gfortran developers asking if compatibility of polymorphic descriptors was possible - I replied that it wasn't, and that there are too many other things that each compiler does differently (I/O, tracking of allocatable variables, different library routines, etc.) to even begin to think about some sort of inter-compiler compatibility.
Your mention of libraries is an important point, but I think the same issue applies to most any language.
Compiler documentation isn't intended to teach you the language, though many use it for that. There are several excellent books that purport to teach you Fortran. Some are better than others. I have not reviewed the more recent entries, but a while ago I did a post about some of them.
Not everyone understood this, and as I write, for a long time the standard didn't help. There was a recent kerfluffle among the gfortran community when the developers fixed a long-time misunderstanding of how BOZ constants should be interpreted. This was the impetus for my post.