Tim Cross <tcross@une.edu.au> wrote:


> To: mmaug@yahoo.com
> Cc: 6686@debbugs.gnu.org; cyd@stupidchicken.com
> Sent: Wed, July 28, 2010 7:51:53 PM
> Subject: Re: bug#6686: 24.0.50; [restore functionality] sql-mode
> 
> Michael Mauger <mmaug@yahoo.com> wrote ...
> >-----  Original Message ----
> >
> >> From: Chong Yidong <cyd@stupidchicken.com>
> >>  To: Michael R. Mauger <mmaug@yahoo.com>
> >> Cc: Tim Cross  <tcross@une.edu.au>; 6686@debbugs.gnu.org
> >> Sent:  Fri, July 23, 2010 12:19:14 PM
> >> Subject: Re: bug#6686: 24.0.50;  [restore functionality] sql-mode
> >> 
> >> Hi  Michael,
> >> 
> >> Could you take a look at this bug  report?   Thanks.
> >> 
> >> 
> >> Tim Cross <tcross@une.edu.au> writes:
> >> 
> >> >  The way sql-mode works in emacs has changed and while  some of the
> >> >  changes are good, there is one major change  that isn't.
> >> >
> >> > In  previous versions of  sql-mode, running queries and executing stored
> >> >   procedures/functions from within the sql interactive buffer did not  
>hold
> >> >  system focus. If the command you executed was going  to take some time 
to
> >> >  run, you could switch to another  buffer/frame and continue working, 
read
> >> >  mail, use w3m  etc. However, this has now changed. When you execute an
> >> >   SQL command or call a stored procedure/function from within the SQLi
> >>  >  buffer, you are effectively locked out from doing anything else in  
>emacs
> >> >  until that command has completed. I'm guessing, but  suspect this change
> >> >  is due to the additional output  processing that has been added to
> >> >  sql-mode, such as  translation of tabs to spaces etc. I have disabled as
> >> >   many features as I can, but this has made no difference.
> >>  >
> >> > I am  using sql-mode to access an Oracle database  using sqlplus.
> >> > Everything  works as it always has with  emacs 23 and this change only
> >> > occured in  emacs 24 either  late last year or early this year.
> >> >
> >> > Could  we  either have the old behavior restored or if that is not
> >>  > possible and if  the change is due to other new features, have a way  of
> >> > turning the new  features off and restoring the old  behavior. I really
> >> > want to continue  using sql-mode and  *really* don't want to have to give
> >> > it up in favor of   something far less pleasant to use than emacs!
> >> >
> >>  >  thanks,
> >> >
> >> > Tim
> >>  >
> >> >
> >> > In GNU Emacs 24.0.50.1   (i686-pc-linux-gnu, GTK+ Version 2.20.1)
> >> >  of 2010-07-19  on  blind-mole
> >> > Windowing system distributor `The X.Org  Foundation', version  
>11.0.10706000
> >> > configured using  `configure   '--prefix=/usr/local''
> >>
> >
> >The only  change that I believe impacted the handling of the command 
>interaction 
>
> >in SQLi mode 
> >
> >is the use of `comint-input-sender' which  may modify the text sent to the 
> >command interpreter
> >before  sending it.  (For example, replacing TAB characters or handling 
> >&placeholders).  This 
> >
> >should not cause the  blocking behavior you describe.
> >
> >I have seen that behavior with  code found on the net that strips out 
> >continuation prompts on 
> >
> >SQL*Plus.  Is that a possible culprit?  
> >
> >To be sure that the input filtering is not the cause you  can:
> >
> >  (sql-set-product-feature 'oracle :input-filter  nil)
> >
> >The other thing you can try is to interrupt emacs under the  debugger when its 
>
> >blocking and
> >look at the backtrace.  
> >
> >Beyond that, more details of your setup of SQL mode might be  helpful.  I'm 
>not 
>
> >seeing the same 
> >
> >behavior but I  don't have sqlplus available to me now.
> >
> 
> OK, I will try the  suggestion of setting input fileter to nil and see if that
> makes a  difference. 
> 
> Note that I do not observe this behavior with emacs 23 and  its version of
> sql-mode. This is on the same system, using the same  configuration and the
> same version of sqlplus. Also note that I do not have  any additional code or
> customizations, other than those provided by sql-mode.  My sql-mode custom
> settings are 
> 
>  '(sql-input-ring-file-name  "~/.sql-input")
>  '(sql-oracle-scan-on nil)
>  '(sql-product (quote  oracle))
> 
> I will let you know what I  find.
> 
> thanks
> 
> Tim

Here's another option: attached is the version of sql.el found in Emacs 23.  
Load 

this rather than the E24 version and see if you see the same behavior.  If you 
do
then this is an E24 bug, else I've got some debugging to do...