PGPLOT version 5.1.1
Version 5.1.1 has been tested with the following operating systems and
compilers. Drivers tested include: GI, GL, NU, PP, PS, TT, VT, WD, X2, XM,
XW (but not all combinations of drivers and systems have been tested
- SunOS 4.1.3_U1, Sun Fortran (f77) 1.3.1, GNU C (gcc) 2.7.0 (tested on SPARC 5).
- Solaris 2.5 (SunOS 5.5), Sun Fortran (f77) 3.0.1, Sun C (cc) 3.0.1 (tested on SPARC IPX).
- Solaris 2.5 (SunOS 5.5), Sun Fortran (f77) 3.0.1, GNU C (gcc) 2.7.2 (tested on SPARC IPX, SPARC Ultra-1).
- OpenVMS AXP V6.1, DEC FORTRAN V6.2, DEC C V4.0, DECwindows Motif 1.1 (tested on DEC 3000/M600).
- OpenVMS VAX V6.1, DEC FORTRAN V6.2, DEC C V4.0, DECwindows Motif 1.2 (tested on VAXstation 4000-90).
Changes in Version 5.1.1
All changes are bug fixes or minor improvements. The most notable
bug fixes are:
- PGOPEN, PGBEG: a device specification like '? ' (question mark
with one or more trailing spaces) causes PGOPEN to issue a blank
prompt for device specification. This was an unintended change in
5.1.0, and has been fixed in version 5.1.1.
- PGBEG: in version 5.1.0, the ordering of panels changed from row
order to column order. The way the sign of the NXSUB argument was
interpreted was precisely the opposite of the documented
interpretation (NXSUB > 0 should give row order, and < 0 should
give column order). PGSUBP has always been wrong, but PGBEG acquired
the incorrect behavior in version 5.1.0. Both PGBEG and PGSUBP now
behave as documented.
- The PostScript driver was ignoring environment variables used to
set the paper size. It now recognizes these variables. In addition, it
will accept requests via routine PGPAP to change the paper size, even
if the size requested is larger than the default size. (You can still
set the default size with environment variables PGPLOT_PS_WIDTH and
PGPLOT_PS_HEIGHT.) There are two side-effects of this change: (1) When
PGPAP is used, a portrait-mode graph is placed in the lower left
corner of the paper (offset by the amount specified by environment
variables PGPLOT_PS_HOFFSET and PGPLOT_PS_VOFFSET; a landscape-mode
graph is placed in the same corner of the paper, but in this case it
appears to be the top left corner! (2) When PGPAP is used, the
bounding-box cannot be guessed when the file is opened, so you should
not use the PGPLOT_PS_BBOX environment variable; if you do not set
this variable, a correct bounding box will be written in the file
trailer (see the discussion in the note on the
PostScript printer driver).
List of Changes
- Revised for version 5.1.1.
- Version number changed.
- It now issues a message encouraging the installer to read
the appropriate README file.
- ver511.txt [new file]
- List of changes (this file).
- Bug fixes: PGPLOT_PS_HOFFSET and PGPLOT_PS_VOFFSET were
not decoded correctly (bug introduced in v5.1.0); bounding box
could be incorrect (probably only on systems with non-static
allocation of Fortran variables). Driver now honors all requests
to change the paper size with PGPAP. Optimization: suppressed
attempts to draw zero-length continuation line segments (thanks
to Remko Scharroo).
- This is an alternative to vtdriv.f. It uses VMS-specific
Fortran, but may work better than vtdriv.f on VMS systems.
- xwdriv.c, pgxwin_server.c
- Bug fix: images were displayed incorrectly on (some?) X-servers
with more than 8 bits per pixel.
- Added example of Cyrillic text on page 3.
- grdtyp.f, grpars.f
- Bug fix: improved minmatch routine for device types to allow,
e.g., /HPGL even if /HPGL2 is also an option.
- Bug fix: (this is a support routine for polygon fill with PGPOLY,
etc.) A polygon with one vertex exactly aligned with the edge of
the window was not clipped correctly (thanks to Remko Scharroo
for the bug fix).
- Bug fix and improvement in header comments. V5.1.0 introduced
a bug: a device argument of '? ' with one or more trailing spaces
did not issue the correct prompt.
- Change version to 5.1.1.
- Correction in header comments.
- This routine was interpreting a negative NXSUB argument
incorrectly. According to the documentation, positive NXSUB
should step through the panels in row order, while negative
NXSUB should step through them in column order; but the routine
was interpreting positive NXSUB as column order and negative
as row order. The behavior
has been corrected to match the documentation. (In version 5.1.0,
pgbeg was changed to call pgsubp, thus introducing
this bug in pgbeg as well.)
- The positioning of labels relative to the axis has been
improved; the displacement of labels from the axis should now
be the same as in PGBOX. Problems were most noticeable when
a large character size was requested. (Thanks to Neil Killeen for the
- The routine was ignoring the first row and column of the array
when finding the scale-factor for the vector length. (Thanks to
David Singleton for pointing this out.)
- Revised to allow a concurrent hardcopy device (Dave Crennell).
- Correct typo (Dave Crennell).
- This new directory replaces sys_vp2200. The files are from
- Configuration file for Fujitsu UXP/M, frt FORTRAN compiler
and /usr/ucb/cc compiler.
- Configuration file for Fujitsu UXP/V, frt FORTRAN compiler
and /usr/ucb/cc compiler.
- Added support for compiling the Motif driver (xmdriv) [mcs].
- Added notes on Linux variants and problems.
- g77_elf.conf [new file]
- Configuration file for Linux systems that use ELF binaries.
- Added notes about use of Sun f90 compiler and GNU g77 compiler.
Added note about problem with the ucb version of "ld".
- f90_cc.conf [new file]
- For Solaris f90 1.1 Fortran compiler (from Ricardo Piriz).
- g77_gcc.conf [new file]
- For GNU g77 Fortran compiler (with gcc).
- Directory removed.
- aaaread.me [new file]
- Includes some notes on incompatibilities between versions
of Motif, C compiler, and VMS.
- This has been rewritten: it now treats logical names as
case-insensitive, and uses $TRNLNM instead of obsolete $TRNLOG.
- makedoc.com [new file]
- A DCL command procedure to extract the documentation from
the source code.
Tim Pearson, California Institute of Technology,
Copyright © 1996 California Institute of Technology