Emacs: Error message when sending a region to a sql connection buffer: Wrong type argument: processp, nil

Question

Based on the article here: https://emacsredux.com/blog/2013/06/13/using-emacs-as-a-database-client/ I wish to have a .sql buffer to edit my queries and once prepared send it by (C-c C-r) to an already open sql connection buffer. My database is postgresql. First of all I open a connection by M-x sql-postgres I enter username, database name and the server host accordingly. Everything works pretty well and I can run queries directly in the new psql shell without any problem. For the sake of this example I renamed this psql shell to django_sql by M-x rename-buffer : django_sql

Just as an example, this could be my sql file which I wish to edit and run in Emacs by using (C-c C-r), that is by sending the region to the already open psql buffer (that is, django_sql), the file name is myscript.sql:

select
    t1.first_name,
    t1.last_name,
    t1.salary
from
    hr_employees as t1
where
    t1.department_id = 100;

Being inside myscript.sql, here is I how I proceed:

1. I run M-x sql-set-product and then I typed postgres
2. Afterwards I did M-x sql-set-sqli-buffer and I selected the name of the connection buffer (psql buffer that I named django_sql)
3. Then I simply selected my SQL query in the file myscript.sql and I did (C-c C-r) to send the selected region to psql buffer (django_sql)

But when I run this I got the following error message:

Wrong type  argument: processp, nil

Could you kindly tell me what did I do wrong, it seems that I proceeded exactly as the above article.

---

Edit (01):

As it was asked in the comments, I added the debug message so that it might be helpful:

Debugger entered--Lisp error: (wrong-type-argument processp nil)
  process-buffer(nil)
  sql-input-sender(nil "select\n    t1.first_name,\n    t1.last_name,*\n    t1.salary\nfrom\n    hr_employees as t1\nwhere\n    t1.department_id = 100;")
  sql-send-string("select\n    t1.first_name,\n    t1.last_name,*\n    t1.salary\nfrom\n    hr_employees as t1\nwhere\n    t1.department_id = 100;\n")
  sql-send-region(1 122)
  funcall-interactively(sql-send-region 1 122)
  call-interactively(sql-send-region nil nil)
  command-execute(sql-send-region)

Answer 1

Finally I found the problem although I cannot explain it. The problem as it seems to me, is due to renaming the psql buffer in Emacs, that is, once I have a psql buffer open and connected to the database, if I run M-x rename-buffer, even if I specify the new name later by runing M-x sql-set-sqli-buffer it seems that Emacs loses somehow the link to the psql buffer when it tries to connect it to my sql script buffer.

What I did instead, was to do C-u sql-postgres, this way emacs prints "*SQL: XXX" and I can then type my desired name which will be prefixed with SQL:

Therefore instead of django_sql-01, the buffer name will be *SQL: django_sql-01* Now I'm really a beginner in this field, so either I completely misunderstand the concept or actually Emacs respects and seeks some pattern in terms of naming the SQL connection buffers (that is, buffers containins a psql prompt)

All I know, is that, by using C-u sql-postgres instead of renaming an already open psql buffer later by M-x rename-buffer, everything works pretty well. I even tested a SQL script with two separate psql buffers (of course with two distinct names) and I run different queries in each of them, and each conntected buffer (each time after doing M-x sql-set-sqli-buffer) provided the expected result.

So, although I cannot explain the true concept and I'm not sure about that, I thought that maybe this could at least be helpful for others who may have encountered the same problem.