Google
 

Trailing-Edge - PDP-10 Archives - bb-bt99h-bb - ftn10.d14
There are no other files named ftn10.d14 in the archive.
                 EDIT DESCRIPTIONS FOR FORTRAN-10-V10                           
  
  
                             EDIT 343    FOR FORDDT
  
  
  
  
[SYMPTOM]
  
  
Auopatch-built FORDDT does not reflect true version number.
  
  
  
[DIAGNOSIS]
  
  
Autopatch won't update the version number of a reserved edit number if
it is the only edit for a particular freeze period.
  
  
  
[CURE]
  
  
Force a dummy update.
  
  
  
********************************************************************************
  
  
                             EDIT 4237   FOR FOROTS
  
  
  
  
[SYMPTOM]
  
A system running version 5.1 of TOPS-20 cannot  debug  (with
DDT)  extended  addressing FORTRAN programs.  The setting of
breakpoints will cause the program to die with  an  "Illegal
Instruction"
  
[DIAGNOSIS]
  
When FOROTS starts-up it begins a garbage collection  phase.
It  will set up a bit map to signify whether a page is being
used or not.  Because the count of the number  of  pages  in
the  section  is  off-by-one,  The  last page in the section
doesn't get marked as "in-use".
  
[CURE]
  
Set the number of pages in a section to  the  correct  value
and the last page in the section will get marked correctly.
********************************************************************************
  
  
                             EDIT 4240   FOR FOROTS
  
  
  
  
[SYMPTOM]
  
An "Undefined Global" is given for  a  MACRO  program  which
calls routine ALOG from FORLIB.
  
[DIAGNOSIS]
  
Because of the way the flagger was  implemented  the  symbol
FLGON.   is  undefined when LINKing this program.  This will
only happen for MACRO program because the  FORTRAN  compiler
will put the definition into the .REL file.
  
[CURE]
  
Make a dummy definition of the symbol which  will  be  found
when it hasn't been defined elswhere.
********************************************************************************
  
  
                             EDIT 4241   FOR FOROTS
  
  
  
  
[SYMPTOM]
  
The Math Library does, in some circumstances, not  give  the
correct  name  of  the  math  function in which an error was
detected.
  
[DIAGNOSIS]
  
When an error occurs from a math library function a  routine
is called to check if the caller of the error was in the low
segment, high segment or FOROTS.   If  the  caller  (of  the
error)  is  known  then  the  text of the error message will
contain the name of the routine which called the error.  The
check for the high segment range is done incorrectly.
  
[CURE]
  
Correctly check for  the  high  segment  range.   Perform  a
GETTAB  to  get  the (real) high segment orign not number of
words  in  the  high  segment  less  the  highest  available
address.
********************************************************************************
  
  
                             EDIT 4242   FOR FOROTS
  
  
  
  
[SYMPTOM]
  
     All FORMAT edit descriptors followed by a blank  are  issued  the
warning of default width used when VMS flagging is used.
  
  
[DIAGNOSIS]
  
     Checking the format width for  zero  in  FMTNC  caused  all  edit
descriptors  followed  by a blank to be issued the VMS incompatibility
warning regardless whether the descriptor had an optional width.
  
  
[CURE]
  
     Set up a flag for FORMAT  edit  descriptors  which  can  have  an
optional  width and flag only those when the default is used.  This is
accomplished by edit 4242 to FORFMT.  This edit  is  a  correction  to
edit 4223.
  
********************************************************************************
  
  
                             EDIT 4244   FOR FOROTS
  
  
  
  
[SYMPTOM]
  
Files written by version 10 of FORTRAN always use dump  mode
I/O.  Other programs are unable to tell what format the file
actually is.
  
[DIAGNOSIS]
  
when the file is closed the FDB  contains  information  that
the  file  was  written  in  .IODMP  mode.   This makes life
difficult for programs which look at this field to determine
how to read the file.
  
[CURE]
  
When closing the file do a rename  FILOP.   instead  of  the
simpler  close FILOP..  This will allow us to rename the I/O
mode before it is recorded in the FDB.
********************************************************************************
  
  
                             EDIT 4246   FOR FOROTS
  
  
  
  
[SYMPTOM]
  
Programs die while trying to close non-disk files.
  
[DIAGNOSIS]
  
After installing edit 4244, FOROTS  tries  to  do  a  rename
FILOP.   to  close  the  file (and rename to the correct I/O
mode).  This should not be done on non-disk units.
  
[CURE]
  
Fix edit 4244.  Don't rename non-disk files when closing.
********************************************************************************
  
  
                             EDIT 4247   FOR FOROTS
  
  
  
  
[SYMPTOM]
  
User gets prompted (in dialog mode) for  another  file  name
when there are too many units opened.
  
[DIAGNOSIS]
  
The file open error causes FOROTS to go  into  dialog  mode.
Don't  prompt  the user since another file name is not going
help.
  
[CURE]
  
If we get this error during a  file  open  operation.   Just
abort with a fatal error.
********************************************************************************
  
  
                             EDIT 4250   FOR FOROTS
  
  
  
  
[SYMPTOM]
  
FORTRAN programs can't run detached.
  
[DIAGNOSIS]
  
Whenever a file is opened a check is done to see if this  is
the  controling terminal.  If the job is detached and we ask
for the controling terminal the  job  will  hang  until  one
attaches to it.
  
[CURE]
  
Add an additional check to see if the job is  detached.   If
it is then we don't have a controling terminal and we should
not ask for one.
********************************************************************************
  
  
                             EDIT 4251   FOR FOROTS
  
  
  
  
[SYMPTOM]
  
FILES OPENED FOR INPUT ARE GETTING RENAMED WHEN CLOSING.
  
[DIAGNOSIS]
  
ANOTHER PROBLEM WITH EDIT 4246
  
[CURE]
  
FIX EDIT 4246
********************************************************************************
  
  
                             EDIT 2565   FOR FORTRA
  
  
  
  
[SYMPTOM]
  
     Long continuation lines cause the compiler to loop infinitely.
  
  
[DIAGNOSIS]
  
     After lexica's buffer is full, it starts writing  over  the  next
contiguous memory locations.
  
  
[CURE]
  
     Don't let LINLCURR (pointer to current entry in lexica's  buffer)
ever advance past the end of the buffer.
  
********************************************************************************
  
  
                             EDIT 2567   FOR FORTRA
  
  
  
  
[SYMPTOM]
  
     Array references to adjustably dimensioned arrays are incorrectly
evaluated  when  there is array declaration with one of the dimensions
having lower bounds not equal to one (e.g.   A  (0:J,K))  following  a
declaration  with a dimension having lower bounds equal to one and the
same upper bound (e.g.  B(J,L))
  
  
[DIAGNOSIS]
  
     Dimension table is shared if the upper bounds of a  dimension  of
two  arrays are the same and the lower bounds of the dimension for the
first array is one.  No comparison is  made  to  check  if  the  lower
bounds are the same.
  
  
[CURE]
  
     Only share dimension information if  both  the  lower  and  upper
bounds are the same.
  
********************************************************************************
  
  
                             EDIT 2570   FOR FORTRA
  
  
  
  
[SYMPTOM]
  
     Compiler gets Internal Compiler Error with arrays that have  both
constant and adjustable bounds.
  
  
[DIAGNOSIS]
  
     IDUSECNT field is being set in constant  table  entries  when  it
should only be set in symbol table entries.
  
  
[CURE]
  
     Before changing the IDUSECNT field of a table entry, check if  it
is a symbol table entry for a .Innnn variable.
  
********************************************************************************
  
  
                             EDIT 2572   FOR FORTRA
  
  
  
  
[SYMPTOM]
  
     Incorrect code generated for open, close  or  inquire  specifiers
that  results  from  multiple  specifiers  depending  on  a value in a
register  which  one  of  the  specifiers  smashes.    The   following
specifiers do not realize that this value has been smashed.
  
  
[DIAGNOSIS]
  
     Register allocation for open specifiers done in the wrong order.
  
  
[CURE]
  
     Allocate registers for open  specifiers  in  the  same  order  as
complexity walk and code generation.
  
********************************************************************************
  
  
                             EDIT 2573   FOR FORTRA
  
  
  
  
[SYMPTOM]
  
     An extra error message is  given  if  there  are  exactly  99-100
continuation  lines.   Also,  when  there are more than 19 consecutive
continuation lines, it should be  flagged  as  ANSI-incompatible  when
compiling with /FLAG:ANSI.
  
  
[DIAGNOSIS]
  
     A bug check in LEXICA should be removed since a similar bug check
was  added  to  LEXCLA  in  edit 2565.  Also, more than 19 consecutive
continuation lines is not flagged incompatible under ANSI flagging.
  
  
[CURE]
  
     Remove bug check in LEXICA.  Also, flag more than 19  consecutive
continuation lines as incompatible under /FLAG:ANSI.
  
********************************************************************************
  
  
END OF  FORTRAN-10-V10