5.0.09 Release Notes

 

All Platforms

 

#489

Backed out change in behavior regarding browse lookup prompts being

displayed even if xkeys specified.

 

Added "-DL" flag to *clerk to tell filePro to display the prompts,

even if xkeys are specified.

 

#547

Shrinking an existing memo/blob to zero bytes would corrupt the blob

file. (Note that this is not the same as deleting the object.)

 

#214

If you use ddir/dprodir to empty a file, and have not yet added any

records nor rebuilt indexes, it is possible that *clerk/*report will

still see the indexes as empty if another process has added records

to the file while you are still in your *clerk/*report session.

 

Note that the records will be correctly seen -- it is only the indexes

that may still be thought to be empty. (The symptom is "no matches

found" or failed lookups.)

 

(No task #)

Add a new end-of-line option for printing -- backslash+LF. Mostly,

this is for the RTF printer table, which requires the backslash at

the end of every line for a continuation marker.

 

#566

 

If PFOLDIX=ON and you build a single-key index in dxmaint with a

descending sort, it erroneously built an old-style index, which do

not support descending keys.

 

#561

 

Change dxmaint to prevent building of indexes on BLOB/MEMO fields.

 

#501

 

Repeated use of F6/D/L to display long variable names in *cabe caused

corrupted list to be displayed.

 

#562

 

Add PFFIXNOLOCK=OFF to back out change related to automatically locking

future executions of a lookup which was modified without a "-p".

 

#536

 

DECLARE GLOBAL name(len,type,g) fields were not properly initialized in

quikstart.

 

#252

 

Assignment/concatenation operators ("=", "<", "{", and "&") now work

with memo fields.

 

eg: memofield = memofield & @TD

  field = memofield

  memofield = field

 

#515

 

If you have an array aliased to real fields, and the only modification

to the record in output processing is "CLEAR arrayname", the record

would not be written without an explicit WRITE.

 

#534

 

LOGTEXT would log some filePro debug information into the log file,

in addition to the LOGTEXT items.

 

(No task #)

 

Add MOUSE PATH [ON|OFF] as an alias for CURSOR PATH [ON|OFF]

 

(No task #)

 

Add PFFIXEDLISTSIZE=ON to prevent filePro from shrinking selection

lists. This allows screen readers for the blind to be programmed

with fixed screen locations for such lists.

 

#569

 

CLOSE() did not return zero on success. (Return value was undetermined.)

 

(Windows network only) #570

 

It was possible that, if one process was accessing an index for the

first time and another was reading/writing the head of that index at

the exact same time, the read/write process would get a Windows

locking violation (Windows error #33).

 

(fileProGI) #573

 

filePro will now break up queued BUTTON_OP commands into 50-command

chunks, to prevent any buffer overruns in the fileProGI client.

 

(*nix only)

 

When using the "+ENV+" command syntax in a user menu, it was possible

to get an error that "exec" was not a valid command/filename.

 

(No task #)

 

Add the ability to put "%varname%" and "$varname" in the title of

user menus, and have them displayed at runtime.

 

(No task #)

 

Add PFSHOWROWCOL=OFF to turn off the row/column display in programs

like dscreen, dmoedef, and *cabe. It can confuse screen readers for

the blind, as the numbers are read every time you press a key.

 

(No task #)

 

Add PFINSERTMODE=ON to set insert mode in by default in *cabe/*clerk.

 

(Native windows, network only) #517

 

Native Windows vetwork version did not respect PFDIRFILTER=ON.

 

#576

 

If you executed a record-number lookup with "-p", and the lookup

failed due to the record being deleted, the record was locked

anyway.

 

(No task #)

 

New flags for dxmaint:

 

 -LY/-LN Include/Exclude index from lists.

 -KY/-KN Save / don't save changes without rebuilding.

 

(fileProGI) #479

 

Short selection didn't set date field types, preventing GI from

properly displaying calendars.

 

(fileProGI) #522

 

Under GI, only a single click of the cancel button is required,

even if the server is Unix. However, this would cause problems

if PUSHKEY "[BRKY]" is used. Change so that double-break is

still needed under Unix if they are sent via PUSHKEY to make

text and GI consistent.

 

#508

 

When rebuilding indexes with -r/-ra, the hidden flag was lost.