OpenMotif 2.3 vs. 2.1.30 Argument
I am a bit confused about the versioning of OpenMotif. I noticed that
the OpenGroup still says that 2.1.30 is the official version, as does
IST. Howeer, on MotifZone, there is an OpenMotif 2.3. Now, in reading
around, I noticed something on IST's website that indicates that
OpenGroup does not “approve” of the OpenMotif versions 2.2 and 2.3. I
also checked out the MotifDeveloper site and again found a lot of
references to versions 2.2 and 2.3 being solely ICS releases that are
However, on the OpenGroup's site, I could not find any official
statement as such. Is there some kind of official word on this stuff?
Aaron Hsu <email@example.com>
Re: OpenMotif 2.3 vs. 2.1.30 Argument
Aaron Hsu wrote:[color=blue]
> Hello everone,[/color]
> However, on the OpenGroup's site, I could not find any official
> statement as such. Is there some kind of official word on this stuff?[/color]
I am not aware of any publicly available statement of OpenGroup
about the official version current version of OSF/Motif
Since I follow this topic, I can point out to this references:
- ICS is the official maintainer of OpenMotif, you can see
that by clicking on link "Motif maintenance partner"
So OpenMotif 2.3 is definitely official.
We also all know of the mostly well deserved
criticism of OpenMotif 2.2
And some related posts:
I think the choice boils down to:
- if you have a legacy application in pure maintenance mode,
then use version 2.1.30
- if you need any of the new features (UFT-8, Antialiased fonts,
support of PNG/JPEG, transparent icons), then your only
option is OM 2.3
You can see the antialiased transparent icons (RGBA), which OM 2.3
enables me to implement, in the following screenshots.
As a developer, I need to be pragmatic and go with the path
which allows me continued development. Luckily, my programs
do not kill people on crash, so my choice is easier.
Re: OpenMotif 2.3 vs. 2.1.30 Argument
On 2008-02-24 16:23:31 -0600, arahne <firstname.lastname@example.org> said:
> Aaron Hsu wrote:
>> However, on the OpenGroup's site, I could not find any official
>> statement as such. Is there some kind of official word on this stuff?[/color]
> I am not aware of any publicly available statement of OpenGroup
> about the official version current version of OSF/Motif[/color]
Here is the closest that I could come to an official statement :
The Open Group has indicated to IST that there are a number of
technical issues surrounding Open Motif 2.2 that will cause it to
re-classify this release as 'experimental'. In particular, it has
been pointed out that this release was only ever an OPEN Motif
release which means that its license for use was only ever valid on
Linux and FreeBSD, and on no other operating system. Open Motif
2.1.30 remains the current supported release.
And then, searching some forms (sponsored by IST) , I found:
ICS released 2.2 without consulting any of the Motif partners, not
OSF, IST, Sun, SGI, KLG, INT, Microlink. They released without peer
review, test. Not even the manager in charge of Motif at the OSF was
There are two camps. ICS, which is on its own, and everyone else,
who continue to insist that OSF 2.1.3x is the official version. Sun
and SGI immediately strip 2.2+ off of their linux offerings.
Let us be clear here. ICS's Motif is not open. Its not in source
forge but ruthlessly controlled on their own machines. They have
denied write access to those such as myself who in conjunction with
SGI and OSF tried to resolve the political and technical issues.
Redhat continue to ship in sheer ignorance.
And then I found a bit of a larger rant :
IST's OpenMotif 2.1.3x is strictly compliant with the last official
OSF release. Fully thread safe, and compatible with the major
operating system vendors releases.
ICS's OpenMotif is their own variant which isnt. It succeeds in
breaking all thread safety, API, binary compatability, and most
seriously breaks every Ada binding on the planet.Open Group, on its
release, commissioned a report into the release, found it so
wanting, that all official sanction for it was removed. Redhat in
sheer ignorance continue to ship, but I wouldnt recommend it under
any circumstances. Sun and SGI's Linux offerings explicitly strip
the ICS motif off of the box and replace with an official OSF
We at IST looked at making a 2.3 release in conjunction with SGI.
But we came to the conclusion very fast that the only thing that
really matters isnt bells and whistles to make the toolkit appear
tick-list functionally capable when compared with newer features of
other toolkits, but strict compatability cross platform. Nothing
else matters at all by comparison, and anything which breaks this is
a very serious mistake.
Besides, its entirelty possible to make Motif do all kinds of
things: from simple things like making the text have multiple
coloured selections or providing tooltips, through to UTF-8 font
support and section 508 accessibility can be done via shared library
plug-ins without modifying the Motif 2.1.30 sources one jot.Motif
2.1.30 is a mighty powerful release. It doesnt need new widgets: the
ones it has are perfectly fine and generic, and anything else tends
to be domain specific, which isnt what a GUI toolkit is about. And
since there are plenty of plug-in components that are compatible and
mature that can be had from a variety of sources, anything domain
specific can be downloaded easily enough. Nor does it need two
comboBoxes, two panes, two toggles, ..., despite what ICS may think.
Any additions simply turn the toolkit from a platform-independent
vendor-neutral generic toolkit into one fit for only one specific
purpose and platform, or work with only one proprietry GUI builder.
And this is a huge mistake: if it doesnt work for all application
domains, X-servers, and platforms with a single constant API, nor
work with all third party widget providors and GUI builder vendors,
then Motif is well and truely dead.And if you do use it, sure as
eggs is eggs, you will find yourself up a blind alley tied to one
specific vendor and platform whether you like it or not. Be not
persuaded that Motif needs the additions which ICS have added: I
find them entirely useless, ill conceived, redundant, and missing
the whole point. I cannot stress enough: Motif 2.2 and 2.3 from ICS
are serious mistakes.
And finally :
There are two sets of issues.
The technical ones may or may not have been addressed, although
there are more and more serious things coming out of the woodwork as
time progresses. The trouble is,
OpenMotif isnt, its rigorously controlled by ICS, and was released
without knowledge of the Open Group. It was not a bloodless coup.
The manager at the OSF has lost his job, and poor old andrew josey
who was conned into accepting the release originally was hauled over
the coals and damn nearly sacked twice. The OSF have withdrawn all
support for the release.
A report was commissioned by the OSF, as indicated. Thereafter,
Acting with SGI as mediator and with full knowledge of the OSF, an
attempt was made to resolve differences and to patch up the release.
To the utter astonishment of SGI, ICS refused all write permission
to myself, at the time the only engineer in the whole world
professionally maintaining Motif for any of the major operating
system vendors, in the most insulting terms.
Since then, no attempt has been made at abridgement. This isnt about
Motif, but ICS, who are claiming to all who will listen that this is
an official release and that it is open. It is nothing of the kind.
For technical issues, the release did break all thread safety, and I
do believe some work has been done in this area. But as I say, there
are more and more serious issues surfacing. Nobody who works on the
toolkit seems to have the slightest idea of the real issues or how
to construct the Motif library. For example, the release simply
plonked "new" resources for the EPak widgets slap bang into the
exising Motif 2.1 strinbg table, without any thought for side effect
or order. The result is catastrophic, for at a stroke it renders all
Ada bindings on the planet volatile. I acted as consultant to
Lockheed a year or two ago, following catastrophic failure of a
helicopter controls: they deadlocked on the pilot, and down she went
toweards the drink with 13 souls on board. This had a Motif
interface. I was horrified to discover that they had cross compiled
from Linux a Motif Ada binding into VxWorks without realizing the
side effect of cross-compiling the string tables which now have the
wrong offsets. All those poor people.
Since then, I have no idea what other butchery lurks. But one can
envisage all kinds of issues surrounding the 2.x "tooltip". If you
must plonk static data into a widget reference, it mustnt be the
widget instance itself. The result here is, you have a component
that is instantiated once on a sp[ecific screen and visual, so at a
stroke this is going to break multi-screen operation. Its a
seriously crap widget anyways, and should have been built along the
XmScreen lines. But nobody there is bright or experienced enough to
So carry on with 2.x if you wish, but you are selling your soul to
The official release is Motif 2.1.30 from OSF. It isnt that from
ICS. And if you believe or have been told this is not so, you have
I note here that all of these comments were made after the beta release
of OpenMotif on MotifZone. Now, ICS may be maintaining the Bug database
for Motif, but it seems to me that there are a lot of companies that
have a hand in Motif. IST and ICS are both listed on the "Other
Organizations" pages of the Open Group site , but it says that the
Open Group "does not endorse any of the work performed by these
companies." Now, this would seem to be in contest with one
interpertation of the term "Motif maintenance partner."
I have sent an email to the Open Group about this, but I have not heard
back from them. Now, I assume that the Open Group's page would
logically have some authority on everything, but I find that the FAQ of
their page  indicates that "Version 2.1.30 is the stable production
version" of Motif. Along with this, they state, "At present there is no
organization supporting the collection and integration of Open Motif
enhancements," which seems to conflict with your interpretation of ICS
being the sole maintenance partner for Motif.
Now, IST states that "the OpenMotif (2.2 or 2.3) distribution that is
supplied with many Linux distributions has been classified as
'expiremental' by the Open Group. " IST continues to imply that by
expiremental, the Open Group means to imply that they do not
necessarily indicate the direction in which future Motif Development
ought to tend. So, while ICS seems to have some kind of endorsement
from the Open Group, it does not seem that anywhere I can see evidence
that the ICS has the right to deem some release of Open Motif Stable
and Officical. It seems that the status of OpenMotif 2.3 from what I
can see can only be reliably interpreted to be "code released by an
organization officially affiliated with the development of OpenMotif."
However, I am quite confused by the seeming contradictions of the Open
Group's site and MotifZone. If indeed the Open Group's site is just out
of date, this is a series issue.
> And some related posts:
I have quoted what seems to be the relevant parts above.
> I think the choice boils down to:
> - if you have a legacy application in pure maintenance mode,
> then use version 2.1.30
> - if you need any of the new features (UFT-8, Antialiased fonts,
> support of PNG/JPEG, transparent icons), then your only
> option is OM 2.3[/color]
It would seem that right now this is partialy a choice, but as ,
, and  would seem to suggest, some of the features supposedly
introduced into version 2.3 were already available in the 2.1.30
release. :-S Of course, I have not seen the evidence that everything in
2.3 is possible in 2.1.30, but I would like to see that. Still, what I
don't understand is whether the Open Group has indeed classified 2.3 as
an expiremental release, despite ICS insisting that it is a stable
release, and, if so, whether the Open Group will push development to a
different direction than that of 2.3. If this does happen, I would be
worried that code using 2.3 may in fact prove to be unsupported in a
future release. Additionally, is 2.3 once again a sole effort of ICS,
and is IST simply not willing to work with ICS, or is it ICS who is
unwilling to work with IST? It seems that from the MotifDeveloper
forums, ICS is not cooperating, and produced a different version of 2.3
than the possibilities examined by IST and SGI. Again, I am not sure.
So, this all just leaves me very confused.
> As a developer, I need to be pragmatic and go with the path
> which allows me continued development. Luckily, my programs
> do not kill people on crash, so my choice is easier.[/color]
Indeed, that makes things easier, and I don't have any one who will die
if my code fails, but I do have people who really do need very reliable
software, and it has to be reliable and maintainable over a long period
of time. So far, 2.3 seems to indicate that it is not quite the stable
system that 2.1.30 touts itself as being, and I am worried that with
the apparent factionism, there may be issues with future
maintainability of my code. The reason I use Motif for development is
that it provides reliability and maintainability. It's stable, in other
words. So, I really wish the Open Group would just make a statement on
all this to give some developers like me some indication of what to
expect in the future. :-/
Aaron Hsu <email@example.com>
Re: OpenMotif 2.3 vs. 2.1.30 Argument
Aaron Hsu wrote:
>> - if you have a legacy application in pure maintenance mode,
>> then use version 2.1.30
>> - if you need any of the new features (UFT-8, Antialiased fonts,
>> support of PNG/JPEG, transparent icons), then your only
>> option is OM 2.3[/color]
> It would seem that right now this is partialy a choice, but as , ,
> and  would seem to suggest, some of the features supposedly
> introduced into version 2.3 were already available in the 2.1.30
> release. :-S Of course, I have not seen the evidence that everything in
> 2.3 is possible in 2.1.30, but I would like to see that.[/color]
I can assure you that the 2.3 features I have mentioned are
not possible in 2.1.30.
I think the cited references state that backwards compatibility
is of paramount importance, and that those new features could
be implemented as plugins (probably by another mid-level library).
Maybe some internal work in that direction has been done by IST,
but I have never seen the source or an official public announcement
of such a development by IST.
Version 2.1.30 is the same old Motif from 10 years ago, with some
Version 2.3 contains new features and new development. Anyone who
develops software knows, that a version, which brings new features
is less stable than a pure bug fix release. I wish I could have
both the stability of 2.1.30 and new features of 2.3, but I can't.
On the other hand, developers do fix bugs OM 2.3
So it is gradually getting better. Probably it will not be
as good as 2.1.30, but if crashes are fixed, and new functionality
works properly, it is good enough for me.
As I understand, OM 2.3 is used mainly on Linux and OSX.
I develop Linux applications for the last 10 years, and
I can assure you Linux programming interfaces break all
the time, in the name of the "progress". And that everything
on Linux is experimental and buggy. The latest blow is
Qt3 -> Qt4, which is totally incompatible. We are not
talking about a recompile, like from 2.1 to 2.2 or 2.3,
we have missing functionality, no automatic migration tool,
practically a rewrite. The teenage "I can't understand it,
let's rewrite" approach is all over Linux.
So OM 2.3, by Linux standards, is both bug free and compatible
with earlier releases - since only a recompile is needed,
with no source changes.
> So, I
> really wish the Open Group would just make a statement on all this to
> give some developers like me some indication of what to expect in the
> future. :-/[/color]
Good luck with that.
In any case, if IST does make a superior version of Motif
with the new features (as "plugin" or any other way), it is not
difficult to jump ship, the new features only affect about 1% of code.
The OM 2.1.30 / 2.3 schism does look like a fork, but in practice
it is an informal split between stable and development release,
not two really competing branches.
My hunch is that ICS would like to sell the license
for OM 2.3 to the remaining Unix workstation vendors
(Sun, IBM, HPC, SGI?), once it reaches the level of
stability, that those vendors require. This would also
partially explain the persistence of the broken OM
open source license. And this would explain why ICS
doesn't want any IST involvement. I do not know anyone
and this is a pure speculation of mine.
Personally, I would much more prefer OM 2.3 being hosted
on one of the normal open source sites, and that all
competent developers, especially Antony Fountain would
have write access.