The back end and libraries are now up for general testing on a variety of boxes and operating systems. The tarball is usually updated as the web page is. To download and install g95 on unix systems, run the following command (originally from Joost Vandevondele):

wget -O - http://ftp.g95.org/g95-x86-
linux.tgz | tar xvfz -

This will create a directory named 'g95-install' in the current directory. Run (or better yet make an appropriate symbolic link to) ./g95-
install/bin/i686-pc-linux-gnu-
g95 in order to run g95.

2007 Archive

November 30

Doug Cox has built some new windows builds.

November 29

Francois Willot pointed out a crash on invalid do-loops that has been fixed.

November 28

I've completed the setup for building on freebsd. Thanks Bud!

November 21

Jack Howarth has posted a tentative fink package, which hasn't been done in a while.

November 20

Bud Dodson has donated the use of a machine for freebsd builds. I have these about halfway done.

November 19

Doug Cox has build some new windows builds.

November 12

Richard Maine, Kangjun Seo, Peter Knowle and Massimiliano Bonomi pointed out that Apple's new Leopard release for x86 was having some problems with g95. I think I've got that fixed now.

Tim Jenness reported having some trouble compiling for multiple architectures under osx. Long story short, f951 now accepts -arch, but does not do anything with it.

Tom Crane also fixed a bug with the -fonetrip in his recent patch.

I also spent some time revamping the internal build system, it's faster and more robust now.

November 8

Doug Cox has built some new windows builds.

November 6

Philippe Marguinaud sent in a small patch fixing a problem with asynchronous I/O that has been applied.

Harald Anlauf sent in a problem with C interop that has been fixed.

October 27

Doug Cox has built some new windows builds.

October 26

Tom Crane sent a patch to allow real variables in DO-loops. This is a bad idea numerically, but some older codes need it. It must be enabled with -freal-loops.

Doug Cox has built some new windows builds.

October 24

Doug Cox has built some new windows builds.

Shaun Forth sent in a weird SELECT statement that crashed. Fixed.

October 16

Walter Spector pointed out a crash on some illegal code that has been fixed.

October 1

Doug Cox built some new windows builds.

September 25

Steven Winfield pointed out a problem with the SCAN intrinsic that has been fixed.

September 12

John Bray pointed out that the -r4 option still wasn't kicking in properly. Got that fixed now.

John Harper pointed out that IEEE_CLASS wasn't working, got that fixed now.

Paul van Delst posted a message to clf on the public/private problem. Richard Maine finally provided an explanation of what was going on. I've put the constraint in 5.2.3 back in, correctly this time.

Mike Behrisch pointed out a regression with recursive subroutines that has been fixed.

September 11

Paul van Delst and about the fifty other people have reported surprise at the public/private derived type constraint of f95. To simplify my life, I've removed the constraint, just like J3 did in the F2003 standard.

September 10

Francois Jacq sent in a problem with C_F_POINTER that has been fixed.

September 7

Doug Cox has built some new windows builds.

September 6

Victor Prosolin pointed out a problem with initializing optional INTENT(OUT) variables via the -fxxx= options that has been fixed.

August 30

Doug Cox reported adding support for Vista in the windows builds.

August 28

Angelo Graziosi and Zepu Zhang pointed out that having -fzero initialize a common block causes a problem with multiple common blocks in the same file. I've turne that behaviour off now.

Al Greynolds sent in a crash on -freal that has been fixed.

Norm Clerman sent in a bug that I suspect was fixed a couple days ago.

August 27

Massimiliano Bonomi pointed out some confusion between powerpc and x86 osx build that has been fixed.

Jan-Willem Blokland sent in a problem that was a duplicate of the PUBLIC/PRIVATE problem from a couple days ago.

Angelo Graziosi and Zepu Zhang sent in a crash on -fzero and common blocks that has been fixed.

John Bray suggested an -r4 option to keep symmetry with the other -rx options, I've got this implemented now.

Matthias Jerg sent in a crash on interface checking that has been fixed.

August 26

Doug Cox has built some new windows builds.

Alberto Garcia and Harald Anlauf pointed out a similar but different constraint from the ones mentioned a couple days ago. Fixed now.

August 22

Emmanuel Reynier pointed out that the command line scalar initialization options (-finteger, -fzero, etc) didn't work with common block variables. Got that fixed now.

Mimo suggested that INTENT(OUT) parameters be initialized with the scalar initialization options, and I've got this implemented as well now. It should make debugging easier.

Reinhold Bader, Yao-ming Huang, Jan-Willem Blokland, Hans-Wilhelm Bolloni, Michael Richmond and Juan Rodriguez-Carvajal pointed out that a recently added constraint was only for fortran 95, removed in F2003. Got this fixed.

Harald Anlauf pointed out a similar but different constraint, also fixed so that it is only checked with -std=f95 is given.

August 18

Doug Cox has built some new windows builds.

August 15

Matthew Halfant sent in a subtle problem with allocatable arrays inside of large derived types that has been fixed.

August 14

Hans-Wilhelm Bolloni, Vivek Rao, Norm Clerman and Michael Richmond reported one last (I hope) bug with generic interfaces that has been fixed.

Michael Richmond also reported another crash that has been fixed.

August 12

Doug Cox has built some new windows builds.

August 10

Hans-Wilhelm Bolloni sent in a regression with generic procedure accessibility that has been fixed.

August 9

Yao-ming Huang sent a crash on pointer array arguments that has been fixed.

Martien Hulsen and Toby White, it turns out, also reported yesterday's bug on generic procedures some time ago.

August 8

Norm Clerman, Balint Aradi and Tom Epperly sent in a problem with generic procedures that has been fixed.

Martin Buchmann sent me a $20.00 Four Peaks gift card. My favorite beer is the Hefeweizen-- it's so yummy.

July 25

Doug Cox has built some new windows builds.

Dominique d'Humières sent in an exponentiation buglet that has been fixed.

July 24

Jos Jansen identified a place in G formatting where memory was being used uninitialized. Hopefully fixed.

July 19

Doug Cox has built some new windows builds.

July 17

Doug Cox has build some new windows builds.

John Harper pointed out that IEEE_IS_FINITE, IEEE_IS_NEGATIVE and IEEE_IS_NORMAL were not implemented. Got them implemented now.

Michael Richmond sent in a regression involving interfaces that has been fixed.

July 16

Karin Meyer sent in a very subtle problem with the ALLOCATE statement that has been fixed.

Sergey Rychkov sent in an updated Russian translation of the manual that is now up.

July 11

Doug Cox build some new windows builds.

July 10

Karl-Albert Breith sent in a subtle problem with generic interfaces that has been fixed.

Norm Clerman sent in a problem with generic interfaces that strangely vanished when I tried to reproduce it. I think fixing Karl's problem also fixed Norm's.

John Harper sent in a spurious warning that has been suppressed.

July 2

Doug Cox has built some new windows builds.

June 27

Henk Krus pointed out a problem with interface checking in f03gl that was a little too tight-- loosened slightly. f03gl is a fortran 2003 interface to gl, glu and glut libraries, so you can write videogames^H^H^H^H^H^Hinteractive physics simulations with fortran if you want.

Aleksander Donev, one of the authors of f03gl, clarified a problem with binding labels that has been fixed.

June 5

Doug Cox has built some new windows builds.

June 4

Doug Cox has built some new windows builds.

Victor Prosolin and Sergey Rychkov updated the Russian translation of the G95 Manual, and is now up and online.

Sebastien Bardeau sent in a crash when there was a warning on module common blocks that has been fixed.

Nick Moore and I exchanged a couple mails last week. He mailed a check (US $15), for "the next round", which is much appreciated.

Reinhold Bader also sent a note with an obscure problem in it last week. After several mails and consulting Richard Maine, we figured that the code is illegal.

Jos de Kloe sent in a subtle problem with not-a-numbers. I fixed the problem as well as I could, but it turned out that gcc is also broken as far as this particular bug is concerned.

May 29

Wolfgang Dobler suggested a change to make module files more resilient to trivial changes in the code, improving "smart compilation".

John Harper sent in a spurious warning that has been fixed.

Paul Boucher pointed out an problem with kind=16 reals that has been fixed. Quad precision reals aren't implemented yet, but there is an unused stub entry for them that had a bad value. Fixed.

Reinhold Bader and I exchanged a couple mails about some screwy results that turned out to be the result of a screwy program. We're not sure why it is illegal and are consulting the godfather.

May 25

Doug Cox has built some new windows builds.

May 24

Toby White sent a crash involving a subtle specification expression that has been fixed.

Geert Jan van Oldenborgh sent in a subtle addressing problem involving large arrays on 64-bit platforms that has been fixed.?

May 22

Doug Cox has built some new windows builds.

May 20

Norm Clerman sent in a crash that has been fixed.

May 17

Jaroslav Hajek has contributed a Czech translation of the manual.

May 14

Doug Cox has built some new windows builds.

May 10

Dominique d'Humières sent in a regression with generic interfaces with alternate returns that has been fixed.

Doug Cox has built some new windows builds.

May 9

Vivek Rao suggested some better wording for an error message that has been fixed. Vivek also sent in a spurious warning that has been fixed.

John Harper pointed out an unimplemented constraint that has been added.

May 8

Doug Cox has built some new windows builds.

John Harper pointed out a spurious warning that has been fixed.

May 3

Michael Richmond, Dominique d'Humières and maybe Victor Prosolin pointed out the fix for yesterday's illegal code was worse than the original bug, so I removed yesterday's patch, pending a better solution.

Daniel Borchmann sent in a crash on abstract interfaces that has been fixed.

Doug Cox build some new windows builds.

May 2

Reinhold Bader pointed out a spurious warning that has been silenced.

Michael Richmond sent in a piece of illegal code that was being compiled. Got that fixed now.

May 1

Vivek Rao donated US $130.00, basically the cost of running the website for a year. Thanks Vivek!

April 27

Doug Cox has built some new windows builds.

April 26

Doug Cox has built some new windows builds.

Vivek Rao suggested some error checking for LBOUND, UBOUND and SHAPE that has been implemented.

Harald Anlauf pointed out that the g77 LTIME intrinsic hadn't been implemented. Got that added now. Harald also sent in a spurious warning that has been suppressed.

April 25

Doug Cox has built some new windows builds.

Reinhold Bader sent in a C interop regression that has been fixed.

Frank Muldoon and I exchanged several letters trying to track down a very subtle crash. The problem was a simple FORALL statement followed by a second statement on the same line.

April 24

Reinhold Bader pointed out some cases of illegal argument association that have been implemented.

Vivek Rao pointed out a sentence on the web page that hasn't been true for a while now. Deleted.

Doug Cox sent in some fixes to the howto's for mkmf and photran that have been incorporated.

April 20

Doug Cox has built some new windows builds.

April 18

Francois Jacq pointed out a crash on illegal INTERFACE statements that has been fixed.

Udo Grabowski contributed a German version of the manual. It has been available for a while now on the gg95 group, but we've had some mail problems that have hopefully been resolved. Vielen Dank Udo!

Doug Cox has built some new windows builds.

April 17

Doug Cox has built some new windows builds.

Matthias Maischak sent in a regression with dummy procedures that has been fixed.

April 12

Joost Vandevondele pointed out a subtle problem with pointers that has been fixed.

April 11

The g95 website (the IP, really) got switched with that of a church earlier today. This is what happens when you register a name at a different place than your web host. The registration is about to expire anyhow, so I am going to move the registrar when I renew.

Doug Cox has built some new windows builds.

April 10

Andres Mujica sent in a subtle problem with vector subscripts that has been fixed.

Doug Cox has built some new windows builds.

April 9

Dominique d'Humières and Valeriy Petrov pointed out a runtime crash on floating-point intrinsics on big-endian architectures that has been fixed.

Paul Scorer-- My mail to you bounced, but the bug is fixed.

Doug Cox has built some new windows builds.

April 4

Vivek Rao pointed out that ONLY clauses were not being correctly pointed at when error messages were generated. Got this fixed.

Vivek also sent in a regression involving a warning on INTENT(OUT) variables that were not set.

Doug Cox has built some new windows builds.

Doug and Michael Richmond sent in a version update for the mingw instructions that has been applied.

Michael found a memory leak that has been plugged.

April 3

S Kaveh sent in a bug on IMPORT statements as well as a crash that has been fixed.

April 2

Doug Cox has built some new windows builds.

March 30

David Bartel pointed out a typo in the English version of the manual that has been fixed.

Angelo Graziosi pointed out some questionable C code that wasn't being accepted by newer versions of gcc that has been fixed.

Alan Greynolds noticed that the compiler had slowed down by an order of magnitude due to some debugging code that was accidentally left in last night.

Doug Cox has built some new windows builds.

March 29

Doug Cox has built some new windows builds.

Denis Brion pointed out that the makefile for the French version of the manual wasn't quite right, it still referred to the English files. Got this fixed now.

Dominique d'Humières, Michael Richmond, Gethin Williams, Alan Greynolds and Paul Scorer pointed out a regression with exponentiation that has been fixed.

March 27

Dominique d'Humières sent in a few more buglets with exponentiation that have been fixed.

Francisco Fadrique sent in a problem with procedure arguments that has been fixed.

March 21

Doug Cox has built some new windows builds.

March 20

Joost Vandevondele and S. Kaveh pointed out a problem with the IMPORT statement that has been fixed.

Vivek Rao pointed out an out of bounds message during slicing that has been improved.

Dominique d'Humières sent three buglets that have been fixed. Two were problems with character specifications, and one fencepost problem on exponential calculations.

Doug Cox has built some new windows builds.

John Backus, author of the original Fortran, died last Saturday, the 17th.

March 19

Dominique d'Humières pointed out a problem with integer exponents in power calculations that were beyond -HUGE(0) that has been fixed.

Valeriy Pertrov pointed out that the stat() family of functions dropped out of the windows version of the library. This was essentially a typo in the windows configuration introduced recently. I've put it back.

March 14

Takeshi Enomoto contributed a Japanese translation of the manual.

March 13

Siegfried MEUNIER-GUTTIN-CLUZEL contributed a French translation of the manual.

Vivek Rao pointed out an error and a typo in the web documentation that have been fixed.

Doug Cox has built some new windows builds.

March 12

Andres Mujica pointed out a case where compile-time array bounds checking was not happening. Got that fixed now.

Vivek Rao sent in a bug where the name of a file involved in an I/O error wasn't being printed correctly. Got that fixed now.

Vivek and Noam Ganot sent in a spurious warning for operator procedures that were invoked by an operator interface that has been fixed. Vivek sent in a similar bug dealing with generic interfaces that has also been fixed.

 

Two weeks ago, I messed up my mail system pretty good. My ISP was no longer letting me forge outbound mail with the firstinter.net return address, so I had to find an alternative solution. I ended up creating an smtp tunnel to the guys that I host the g95.org website. The last thing I did was to set up a firewall rule so that no one could use the tunnel as an open smtp relay. The system was working perfectly, and I botched adding the rule to the startup scripts and ended up closing the tunnel.

From that point out, outbound mail wasn't going anywhere. It took a week to bounce, and then another week for the bounce to bounce, since I don't have a postmaster here. I started getting the failure notices today. Mail less than a week old should get delivered now, and I am resending mails between a week and a week old.

March 8

Doug Cox, Michael Richmond and Angelo Graziosi reported problems building under windows that has been fixed.

Once the difficulties were clear, Doug build some new windows builds.

March 7

Doug Cox has built some new windows builds. Doug has also collected and updated some documentation on using Photran and mkmf with g95 that is now up.

Vivek Rao sent in a problem with an extension of passing character arrays that has been fixed.

Brian Toby pointed out a couple of missing g77 intrinsics that I've added: gerror(), ierrno(), itime(), and second().

March 6

Jimmy James pointed out a regression with -std=f95 that has been fixed.

Doug Cox has built some new windows builds.

John Harper reports that DATE_AND_TIME() (thankfully) works in New Zealand, even with NZDST in effect.

Bill McLean pointed out some 32-bit relocations within some assembler code that handles kind=10 reals on x86_64 platforms that were preventing the library from being linked as a shared library. I've managed to get rid of these relocations without sacrificing efficiency.

March 5

Andres Mujica found a problem with logical SELECT statements that has been fixed.

Doug Cox has built some new windows builds.

March 4

Michael Richmond and John Harper sent in a regressions with DATE_AND_TIME() that has been fixed. I'm really hoping I've got it right now.

Michael also sent in a traceback regression that has been fixed.

Vittorio Zecca sent in a subtle regression with statement functions that has been fixed.

Vivek Rao suggested warning about private module procedures that are never invoked. I've expanded the -Wunused-module-procs to include these cases as well.

Doug Cox has built some new windows builds.

March 3

David Davies pointed out a problem with SYSTEM_CLOCK() that has been fixed.

Tobias Burnus sent in a crash on C interop that has been fixed.

March 1

Michael Richmond pointed out that the problem with GMT offsets in DATE_AND_TIME() was a little more difficult than anticipated. I've got a much better solution now.

Doug Cox has built some new windows builds.

February 28

Vivek Rao sent in a crash in the PACK() intrinsic that has been fixed.

Bill McLean pointed out a subtle linking problem on x86 using -fPIC that has been fixed.

John Harper pointed out a problem with DATE_AND_TIME() that was broken on solaris. I've got a more portable version in place now.

February 27

Michael Richmond send in a regression in function declarations that has been fixed as well as a regression on dummy procedures that has been fixed.

Doug Cox has built new windows builds.

February 26

Joe Krahn pointed out a spurious warning that has been fixed.

Kris Kuhlman pointed out a somewhat misleading warning that has been made more clear.

February 23

Tom Crane reported that version 7.2 of the WIEN density functional code now works with g95.

I was convinced that calling variable argument functions from C was illegal, but have decided to offer this as an extension.

February 22

Got into an interesting discussion on the J3 list today. Seems the standard has been fixed so that you can't call C subroutines with a variable number of arguments, like, say, printf(). This seemed silly to me and it led to a bunch of messages.

February 21

Dominique d'Humières and Michael Richmond pointed out a regression in derived types that has been fixed.

Doug Cox has built some new windows builds.

February 20

Angelo Graziosi pointed out a problem with the web page that has been fixed.

Jimmy James sent in a crash on misusing PROGRAM statements that has been fixed.

Joost Vandevondele sent in a subtle code that required an error. Got it erroring out now.

Nuno Pinhão found an equally subtle problem in reading namelists that has been fixed.

February 18

Sourceforge has been unable to process any web updates for about a week now. If you're reading this now, it's because the DNS updates have propagated to your machine and you're reading this on the commercial hosting service. The old address, g95.sf.net will remain, hopefully in sync, but www.g95.org should be considered the primary.

Dan Davison pointed out a bug in the INSTALL document that has been fixed.

February 16

Doug Cox has built some new windows builds.

February 15

David Davies pointed out a problem with WHERE statements that has been fixed.

Doug Cox has built some new windows builds.

Vivek Rao pointed out a regression with allocatable arrays that has been fixed.

February 14

Lionel Guez pointed out a regression in the MERGE() intrinsic that has been fixed.

Reinhold Bader sent in a pair of regressions for procedure pointers that have been fixed.

February 9

Manuel Guidon sent in a crash on the C_F_POINTER() intrinsic that has been fixed.

Stephen Bespalko sent in a spurious warning that has been suppressed.

Doug Cox has built new windows builds.

February 8

Joost Vandevondele pointed out a pair of crashes on procedure pointers that have been fixed.

Doug Cox has built some new windows builds.

February 7

Harald Anlauf sent in a crash on generic procedures that has been fixed.

Pierre Hily-Blant sent in a problem with the WHERE statement on 64-bit platforms that has been fixed.

Robert Funnell sent in a problem with tracebacks that has been fixed.

February 6

Doug Cox has built some new windows builds.

February 5

Sergey Korop pointed out a subtle problem with nonstandard min and max statements that has been fixed.

Michael Richmond, Dominique d'Humières, Doug Cox, Pablo Rios, Erik Kruyt, John Dormand (probably), Kris Kuhlman (probably) and Al Greynolds pointed out a nasty regression with character functions that has been fixed.

Bill McLean contributed a g95 spec file for building RPMs. It is available in the usual download area.

February 2

David Davies pointed out an obscure crash (you had to use a couple of options) that has been fixed.

Al Greynolds sent in a regression of the PACK() intrinsic related to the conformance checking added recently that has been fixed.

February 1

Richard Hartman identified a regression in external functions that has been fixed.

Victor Prosolin has translated the g95 manual into Russian. I'd be happy to have the manual in other languages as well. French, German and Italian would be my preferences, since g95 is heavily downloaded from these places.

January 31

Andrea Calaon pointed out a small mistake in the Photran instructions that has been fixed.

Doug Cox has built some new windows builds.

January 30

George Bohannon pointed out an optimization when passing arrays as actual arguments that has been implemented.

Joost Vandevondele pointed out an unimplemented constraint that has been added for strict f95 mode.

Paul Suckling pointed out some updated instructions that appeared on the Photran mailing list for using Photran with g95. Doug Cox took the post by Hani and converted into html which has been added to the howto page.

January 29

Vivek Rao pointed out a mispelling in an error message that has been fixed.

January 24

John Bray sent in a spurious warning that has been fixed.

Stephen Bespalko sent in a subtle problem parsing substrings that has been fixed.

Clayton Evans pointed out a problem with the stat() extension under windows that has hopefully been fixed.

Doug Cox sent some new text for mingw installation instructions that has been added to the manual.

January 23

Exchanged a couple letters with Michael Richmond, Doug Cox and Skip Knoble fiddling with memory allocation under mingw/windows.

Doug Cox has built some new windows builds.

January 21

Michael Richmond pointed out a regression in structure constructors that has been fixed, as well as a dozen places where error pointers weren't quite pointing to the error that have been fixed.

Tobias Burnus sent in a problem with USE statements that has been fixed.

Mark Herndon pointed out that STOP with numeric and character stop code weren't behaving quite the same. I've fixed that.

Joost Vandevondele sent in a subtle problem with generic interfaces that has been fixed.

Vivek Rao suggested a conformance test on the PACK() intrinsic that has been implemented.

January 18

Michael Richmond pointed out a pair of unreported errors that have been fixed.

January 17

Vivek Rao sent in a crash under -Wall that has been fixed.

James Spencer sent in a structure constructor crash that has been fixed.

Stefan Wippermann-- my reply to you bounced. The gist of my answer was to reconfigure g95 and recompile.

January 16

Dominique d'Humières pointed out another problem with ISHFTC that has been fixed. Dominique also pointed out a problem reading modules that has been fixed.

Kris Kuhlman pointed out a sign error in the kind=10 SINH() function that has been fixed.

Doug Cox has built some new windows builds.

January 15

Dominique d'Humières, Michael Richmond, Doug Cox and Michael Caracotsios pointed out a regression from format strings that has been fixed.

Joost Vandevondele pointed out a regression in C_LOC() of structure members that has been fixed.

Reinhold Bader also pointed out an F2003-ism regarding pointer assignment that was not trapped, but not implemented. I've g95 to issue an error for now. Reinhold also sent in a spurious error on PROTECTED symbols that has been fixed.

Dominique d'Humières pointed out a regression in ISHFTC() that has been fixed.

Delbert Franz pointed out some sloth in the printing of floating point numbers. I found an optimization that sped things up by about 40%.

January 13

Doug Cox has built some new windows builds.

January 12

Michael Richmond pointed out that spaces in format strings were significant. Got that fixed now.

Doug Cox found a tricky problem with ISHFTC that has been fixed.

January 11

Doug Cox has built some new windows builds.

January 10

Doug Cox has built some new windows builds.

Reinhold Bader sent in a crash and a bug on C interop issues that have been fixed.

Vittorio Zecca pointed out that a quiet not-a-number was being used for G95_MEM_INIT=nan. I've changed it to a signalling nan.

January 9

Joost Vandevondele, Michael Richmond and Harald Anlauf pointed out a regression in target expressions that has been fixed.

Reinhold Bader found another C-interop related crash that has been fixed.

Doug Cox has built some new windows builds.

Reinhold Bader pointed out another bug with function pointers that has been fixed.

Janaury 8

Martin Dix sent a regression with pointer expressions that has been fixed.

Bill Paxton pointed out a spurious warning that has been fixed.

Doug Cox has built some new windows builds.

Reinhold Bader sent in a pair of problems with F2003 C interop that have been fixed.

Vittorio Zecca sent in a pair of bugs that have been fixed.

January 5

Joost pointed out that his crash on procedure pointers from yesterday wasn't quite fixed. Got this fixed now.

Dominique d'Humières pointed out an invalid error in the ISHFTC() intrinsic that has been fixed.

Alex Brincko pointed out that COMMON variables weren't being emitted as part of debug info. G77 had the same problem, and was the cause of me getting involved with fortran compilers. Now fixed in g95 as well as g77.

January 4

Doug Cox has built some new windows builds.

Michael Richmond found some regressions from last night's fixes to error messages that have been fixed.

Joost Vandevondele and John Harper forwarded an interesting discussion from comp.lang.fortran regarding some aliasing issues. An optimization for array arguments has been implemented.

Joost also sent in a problem with optimization of C_PTR's under optimization. Doug did some testing against the 4.1 build, and it turned out the problem was generic to g95. I've gone back and rewritten how g95 represents C_PTR's internally, again. Things seem to work now.

Vivek Rao and Skip Knoble pointed out that the F2003 IEEE support was nonexistent. I've added some initial support for it now.

January 3

Luis de Vries pointed out a problem with smart compilation that has been fixed.

Joost Vandevondele found a subtle problem with procedure pointers that has been fixed.

Reinhold Bader sent in a less-subtle problem with procedures pointers that has also been fixed.

Franciso Fadrique sent in a regression involving pointer expressions that has been fixed.

Michael Richmond pointed out an error message that incorrectly changed when the order of program units in the program changed.