Cross gcc?

[system]

Dan Haynes <haynesd@ibm.net> wrote:

pete@qnx.com > wrote:


Dear Mr. Haynes (or anyone else who’s interested)

If you want our modified source to gcc, send email to > cburgess@qnx.com> ,
and he will tell you where you can get it from.

This is what I was expecting to read - not (paraphrasing) “It’s not
availble, talk to QSSL laywers”.

Well that response did not come from a QSSL employee.

In every other experience I’ve had with GNU software there have always been
sources (and often a binary) readily available on an FTP server or an FAQ
explaining where/how to request/obtain the sources/binaries. QSSL is an
exception to that. There appears to be no such information about the tools in
the FAQs (or if it’s there, it is buried deeply enough to be difficult to find
using the provided search engine) and apparently nothing on the FTP site. A
search of the entire > www.qnx.com > site for ‘gcc compiler sources’ produces no
relevant information and no links to relevant information.

Valid criticisms all. Thanks for pointing them out. As you probably know,
Colin has remedied the problem with due haste.

When no sources are
mentioned, no contact information provided, no posts in the any of the
qdn.public.* heirachy mention them, and a search using Google and AltaVista
for “qnx compiler sources” produces nothing useful anywhere on the net and a
post to the news groups yields what appears to be a serious answer from a
knowledgable source that says (again paraphrasing) “It’s not availble, talk to
QSSL laywers”, it’s pretty reasonable to assume something really stupid is
going on.

You may not realize that QDN and QUICS are technical forums. For legal
questions or business questions, you’re best advised to ask your sales
contact directly. You do have a contact here like an FAE or a sales rep or
someone in Alliances, don’t you?

And I would take issue with your statement that it’s reasonable to assume
something stupid is going on because someone who doesn’t work here posted
a response telling you to contact our lawyers. It ignores the fact that
people who do work here also posted telling you how to get the source
in a timely manner.

So now that we’ve had fun keeping the VSIK (very special inside knowledge) a
secret before finally let loose with a link, I can now get on with my job.

Considering that even a notoriously proprietary outfit like Microsoft readily
offers the sources for the Windows CE kernel just for the cost of shipping the
CD, one would think that finding the sources to a port of free development
tools would have required a good bit less searching and waiting.

Sorry you think you had to wait a long time, but you got your answer the
next business day, and I assume you now have the gcc source. It may have
seemed like a long wait to you though since it appears your original post
was made Friday night after business hours.

[Mario_Charest1]

I was not expecting a post saying effectively “It’s not available, talk to
their lawyers”. Since that information came from a knowledgable source who
posts frequently
and seems a reasonable guy, I took the whole message literally rather than
partially tongue
in cheek the way it was apparently intended.

I take the blame for that and appologize.

And stating that the FSF’s s/w is systematically of lower quality is at
worst BS,
and at best a troll. I’d suggest comp.os.windows.advocacy if you wish to
make

a case that GP s/w is inherently of inferior quality to receptive ears.

Most users of Unix-like systems have derived incredible value from GPLed
s/w,

and even if some aren’t die-hard fans of the FSF’s politics, acknowledge
this.

I’ve had a different experience. Having used good quality commerical tools
for the last 15
years, the last couple of years dealing with “oh you need to get this
patch which you would
have known had you spent time monitoring 20 different mailing lists”, or
“oh you need to
download that from here and get this from there… they’ve never been
tested together but
they should work…” bit has been very frustrating for me. I’ve spent
most of the last 13
years working on systems with extremely strict government regulations. The
last two years
I’ve been on a project using “free” software. It has been more of a
hinderance than a help
to me due to completely uncontrolled and/or poorly tested releases of
various packages. That
is a fact, not b.s. or a troll.

Anyway, I appreciate your help with finding the compiler sources. Thank
you!

Now let’s see what bit of ancient Unix history I have to learn so I can
mis-configure that
build… >

If it’s QNX4, I’m pretty sure it was done with Watcom.
Uou could instead get your hand on Watcom (Symantec as nothing to
do with it), either the native QNX4 or Windows version. People have used
the Windows version of Watcom to build QNX4 executable.

Watcom is now open source, so you could also get your hands on that.

I think you should review your opinion about “I’m just now realizing that it
seems some
people actually develop under the target system. In my case this is not
desirable or practical”.
I do not agree, it is very desirable and practical… It’s not “some
people” it’s “almost everybody”
develop under the target.

You don’t have to give up your environment (Windows). All you need is to
install a network card on the target, and setup another machine that
will act as a QNX4 server. So you edit your files on windows and with
Phindows you open up a shell on the QNX4 server and run make.

You could even run/debug the application on the server (assuming
it doesn’t required specific hardware).

The tools are not as fancy as you may be used to, but the QNX4 environment
provide the tools by itself. You just have to learn it. It can do
incredible
think that can make the development cycle very efficient.

Consider forgetting about compiling on Windows, give a try to setup
a QNX4 server and let it do the compilation, that you initiate from Windows.

Dan Haynes

[Colin_Burgess1]

Hmmmm, here’s what I found on

http://www.qnx.com/cgi-bin/dir_find.cgi?/usr/free/

…
Special Notes:

\

1. QNX Software Systems is committed to providing the source code for all
   GNU utilities (and other software subject to the GPL) which it
   distributes. Such utilities include awk, rcs, gmake, gzip, and sed. If
   you can’t find the GNU source code you are looking for under the free
   software area, please contact freesoft@qnx.com and we will make a copy
   available.
   …

Dan Haynes <haynesd@ibm.net> wrote:

Sam Roberts wrote:

Previously, Dan Haynes wrote in qdn.public.qnx4:

Getting back to my original question, if someone has a pointer to the sources for the
compiler port(s), I’d appreciate it. Since the GPL clearly states that the GPL applies
to each and every recipient of the GPL’ed sources (i.e. anyone who receives the
sources has a full license to re-distribute them) I assume someone (hopefully someone
who has a good bit more bran in their diet) already has posted it on a web or ftp site
somewhere.

For which O/S? Assuming you can read the name of the the newsgroup you are posting
in, then gcc for QNX4 is not distributed by QSSL, but the port was supported to
some extent by them, see: > www.teaser.fr/~jcmichot

Thank you for the info!

I did not know that there were different ports of gcc - I assumed there was one for both
systems since it sounds like QNX and Neutrino are very similar.

I was basically handed a license floppy, a CD, and terminal hardware with most of the
instructions written in German. Now I need to figure out what it will take to make something
happen with it. I fiddled around with QNX many, many years ago but just basic
evaluation/experimenting. I am just now in the beginning stages of learning what it is like
today.

For RTP, gcc src is not posted (the GPL does not require this), but those who have
asked on the beta groups have received, afaik, as required by GPL. Note that if
QSSL hasn’t distributed gcc for RTP to you, then they are under no obligation to
distribute the source to you either. You could ask Igor or Mario, they have the
right to distibute the source to you if they wished.


Note: the GPL applies to “distributed” s/w. There is precedent (Corel’s Linux
distribution) for this to mean that the source does not have to be made
available during early or limited beta testing. The RTP clearly is still in this
stage. The justification is that distributing and packaging of s/w is a burden
that can be left until release, and not be done at every minor upgrade. You
may not agree with this, and it has not been tested in court, but the Linux
hordes failed to massacre Corel on this, so I’d say its been tested as well as
the GPL has ever been tested.

I don’t have any idea what the ‘RTP’ package is other than some marketing blurbs on the web
page. It doesn’t look like it exists yet, so it’s not of interest to me. My limited (and
perhaps flawed) understanding is that since the Watcom/Symantec compiler is belly up, gcc
was tapped as the ‘official’ development compiler for QNX/Neutrino. I made the assumption
that little (if any) development would be done on the target platform and I was (and still
am) a bit confused as to why the tools don’t simply come set up as cross-development suite
hosted on NT/Linux/FreeBSD.

I’m just now realizing that it seems some people actually develop under the target system.
In my case this is not desirable or practical. I just want to build an executable from tools
hosted on an ordinary NT workstation, copy the output file(s) to a PCMCIA card and plug that
into the target. I really don’t want the application programmers to have to learn Unix/QNX
trivia just to check out the sources and build the project.

Anyway, in reading the previous posts to see if I could find the sources, I saw posts from
QSSL employees. I made the assumption that if I posted here I would simply get a pointer
from someone there to ‘e-mail so and so and they’ll tell you how to get it’ or ‘it’s on the
ftp site at…’. I was not expecting a post saying effectively “It’s not available, talk to
their lawyers”. Since that information came from a knowledgable source who posts frequently
and seems a reasonable guy, I took the whole message literally rather than partially tongue
in cheek the way it was apparently intended.


And stating that the FSF’s s/w is systematically of lower quality is at worst BS,
and at best a troll. I’d suggest comp.os.windows.advocacy if you wish to make

a case that GP s/w is inherently of inferior quality to receptive ears.

Most users of Unix-like systems have derived incredible value from GPLed s/w,

and even if some aren’t die-hard fans of the FSF’s politics, acknowledge this.

I’ve had a different experience. Having used good quality commerical tools for the last 15
years, the last couple of years dealing with “oh you need to get this patch which you would
have known had you spent time monitoring 20 different mailing lists”, or “oh you need to
download that from here and get this from there… they’ve never been tested together but
they should work…” bit has been very frustrating for me. I’ve spent most of the last 13
years working on systems with extremely strict government regulations. The last two years
I’ve been on a project using “free” software. It has been more of a hinderance than a help
to me due to completely uncontrolled and/or poorly tested releases of various packages. That
is a fact, not b.s. or a troll.

Anyway, I appreciate your help with finding the compiler sources. Thank you!

Now let’s see what bit of ancient Unix history I have to learn so I can mis-configure that
build… >

Regards,

Dan Haynes

–
cburgess@qnx.com

[Dan_Haynes]

In every other experience I’ve had with GNU software there have always been
sources (and often a binary) readily available on an FTP server or an FAQ
explaining where/how to request/obtain the sources/binaries. QSSL is an
exception to that. There appears to be no such information about the tools in
the FAQs (or if it’s there, it is buried deeply enough to be difficult to find
using the provided search engine) and apparently nothing on the FTP site. A
search of the entire > www.qnx.com > site for ‘gcc compiler sources’ produces no
relevant information and no links to relevant information.

Valid criticisms all. Thanks for pointing them out. As you probably know,
Colin has remedied the problem with due haste.

Much appreciated by whoever comes along after me, I’m sure. Thank you.

When no sources are
mentioned, no contact information provided, no posts in the any of the
qdn.public.* heirachy mention them, and a search using Google and AltaVista
for “qnx compiler sources” produces nothing useful anywhere on the net and a
post to the news groups yields what appears to be a serious answer from a
knowledgable source that says (again paraphrasing) “It’s not availble, talk to
QSSL laywers”, it’s pretty reasonable to assume something really stupid is
going on.

You may not realize that QDN and QUICS are technical forums. For legal
questions or business questions, you’re best advised to ask your sales
contact directly. You do have a contact here like an FAE or a sales rep or
someone in Alliances, don’t you?

I’m sure there probably is a contact but it’s probably through Wincor/Nixdorf.
I did not post any legal questions, I was merely responding/trying to get more
information on my request for pointers to the compiler sources.

And I would take issue with your statement that it’s reasonable to assume
something stupid is going on because someone who doesn’t work here posted
a response telling you to contact our lawyers. It ignores the fact that
people who do work here also posted telling you how to get the source
in a timely manner.

As you can see in my original text, it was point a. plus point b. plus point
c. plus the fact that I didn’t see any refutation. Simply lack of a quick
response was not the only information I made my (incorrect) assumption on.

So now that we’ve had fun keeping the VSIK (very special inside knowledge) a
secret before finally let loose with a link, I can now get on with my job.

Considering that even a notoriously proprietary outfit like Microsoft readily
offers the sources for the Windows CE kernel just for the cost of shipping the
CD, one would think that finding the sources to a port of free development
tools would have required a good bit less searching and waiting.

Sorry you think you had to wait a long time, but you got your answer the
next business day, and I assume you now have the gcc source. It may have
seemed like a long wait to you though since it appears your original post
was made Friday night after business hours.

QSSL did respond fairly quickly, and I appreciate that. Still, waiting several
days for something that is typically publically posted/readily available is
too long. Sunday here is Monday in some places in the world. We’ve got people
trying to make semi-rational decisions on my “best guesses” that are based on
what I can find on the web or through reading/experimenting. I dislike that
(it’s not engineering it’s guessing) but it’s the way the business world goes
and it’s annoying when I have to dig for something so trivial.

I incorrectly assumed that QSSL was not making the FSF sources available, and
that is definitely not the case. The information was not as easy to find as it
should have been, that has been corrected. I consider the issue
resolved/closed.

My apologies for my error, and a kick in the shin to QNX for not having the
information easily available to start with.

Regards,

Dan

[Mario_Charest1]

My apologies for my error,

And my apologies too, for turning this into a legal issue.
That wasn’t my intention, but it was my fault

[Dan_Haynes]

Mario Charest wrote:

I was not expecting a post saying effectively “It’s not available, talk to
their lawyers”. Since that information came from a knowledgable source who
posts frequently
and seems a reasonable guy, I took the whole message literally rather than
partially tongue
in cheek the way it was apparently intended.

I take the blame for that and appologize.

Thanks, but there is no need to apologize, I’ve been around newsgroups long
enough that I should have asked for clarification first. It just ‘fit’ with my
experiences leading up to that and seemed quite possible.

A little smiley injected here and there wouldn’t hurt though!

If it’s QNX4, I’m pretty sure it was done with Watcom.
Uou could instead get your hand on Watcom (Symantec as nothing to
do with it), either the native QNX4 or Windows version. People have used
the Windows version of Watcom to build QNX4 executable.

Watcom is now open source, so you could also get your hands on that.

Open sourced Watcom was still vapor when I last checked (a week or two ago),
and from what I read on the newsgroups I got the impression that when it was
released it would not include the QNX target. I’m guessing/assuming that that
development was a seperate effort by QSSL and therefore not a part of the
regular Watcom development…?

Is the QNX target expected to be part of the open sourced Watcom?

I think you should review your opinion about “I’m just now realizing that it seems some
people actually develop under the target system. In my case this is not
desirable or practical”.
I do not agree, it is very desirable and practical… It’s not “some
people” it’s “almost everybody”
develop under the target.

You don’t have to give up your environment (Windows). All you need is to
install a network card on the target, and setup another machine that
will act as a QNX4 server. So you edit your files on windows and with
Phindows you open up a shell on the QNX4 server and run make.

You could even run/debug the application on the server (assuming
it doesn’t required specific hardware).

Multiple developers (around 20) and probably not enough hardware to provide
one of every possible variation of hardware for each developer. Development
targets will likely be in a lab seperate from the developers’ offices with
multiple users testing changes on a single machine (i.e.
stand-in-line-isn’t-this-as-much-fun-as-batching), and the targets will likely
need to be rebooted to load the new software, interrupting anything that might
be going on at the time.

So I’d really rather not have folks building and running on the same box.

The tools are not as fancy as you may be used to, but the QNX4 environment
provide the tools by itself. You just have to learn it. It can do
incredible think that can make the development cycle very efficient.

I don’t doubt that QNX is quite capable, the point is that we already have an
ANSI C source tree, build procedures and a lot of investment in those. I want
to limit the exposure to the QNX system details to a couple of folks who will
know all about it, and keep everyone else blissfully ignorant of what happens
inside the magic box. The application shouldn’t care what’s under the hood as
long as the operating system abstractions look the same to it. (at this point
a grand piano falls from a height of a thousand metres and destroys most of
that ideal world, but that’s the goal anyway)

Consider forgetting about compiling on Windows, give a try to setup
a QNX4 server and let it do the compilation, that you initiate from Windows.

Compiling on QNX4 might be an option (perhaps even a neccesity) but my
preference is to keep our existing build, swap out the GNUPro 68k
cross-compiler for the QNX cross-compiler and call it done. Then I can spend
the rest of the day spreading vicious rumours via the newsgroups!

Thank you for the input!

Regards,

Dan Haynes

[system]

Dan Haynes <haynesd@ibm.net> wrote:

I was not expecting a post saying effectively “It’s not available, talk to
their lawyers”. Since that information came from a knowledgable source who
posts frequently
and seems a reasonable guy, I took the whole message literally rather than
partially tongue
in cheek the way it was apparently intended.

I take the blame for that and appologize.

Thanks, but there is no need to apologize, I’ve been around newsgroups long
enough that I should have asked for clarification first. It just ‘fit’ with my
experiences leading up to that and seemed quite possible.

A little smiley injected here and there wouldn’t hurt though! >

Oh well… I apologize too. I hate being left out

Open sourced Watcom was still vapor when I last checked (a week or two ago),
and from what I read on the newsgroups I got the impression that when it was
released it would not include the QNX target. I’m guessing/assuming that that
development was a seperate effort by QSSL and therefore not a part of the
regular Watcom development…?

Is the QNX target expected to be part of the open sourced Watcom?

For official word on this, you’ll all have to wait for an official
statement from our corporate communications dept. However, we have
in the past officially stated that we support gcc for NTO development,
so that is a known issue.

Also, our official compiler for QNX4 is still Watcom 10.6, and we still
have to officially charge money for it.

When and how this will change is unknown at this point, and any official
word will not likely be disseminated through these newsgroups.

So I’d really rather not have folks building and running on the same box.

Understood. I think what Mario is getting at is that most people develop
on one box that runs QNX or NTO, and target another box running QNX
or NTO.

I don’t doubt that QNX is quite capable, the point is that we already have an
ANSI C source tree, build procedures and a lot of investment in those. I want
to limit the exposure to the QNX system details to a couple of folks who will
know all about it, and keep everyone else blissfully ignorant of what happens
inside the magic box. The application shouldn’t care what’s under the hood as
long as the operating system abstractions look the same to it. (at this point
a grand piano falls from a height of a thousand metres and destroys most of
that ideal world, but that’s the goal anyway)

Understood. You want to cross develop, and let your developers stay on
the systems they already know.

[Igor_Kovalenko2]

The bottom line, he’ll have to buy Watcom 10.6 from QSSL anyway, to get
libraries/headers. Then use either GCC or Watcom 11 on Windows to
cross-compile. I don’t know if it is possible to link QNX4 executables
on Windows with Watcom.

Unless of course, Metrowerks will come up with full cross-development
system for both QNX4 and NTO. They currently only support NTO and I
started to doubt promices about QNX4

• igor

pete@qnx.com wrote:

Dan Haynes <> haynesd@ibm.net> > wrote:
I was not expecting a post saying effectively “It’s not available, talk to
their lawyers”. Since that information came from a knowledgable source who
posts frequently
and seems a reasonable guy, I took the whole message literally rather than
partially tongue
in cheek the way it was apparently intended.

I take the blame for that and appologize.

Thanks, but there is no need to apologize, I’ve been around newsgroups long
enough that I should have asked for clarification first. It just ‘fit’ with my
experiences leading up to that and seemed quite possible.

A little smiley injected here and there wouldn’t hurt though! >

Oh well… I apologize too. I hate being left out >

Open sourced Watcom was still vapor when I last checked (a week or two ago),
and from what I read on the newsgroups I got the impression that when it was
released it would not include the QNX target. I’m guessing/assuming that that
development was a seperate effort by QSSL and therefore not a part of the
regular Watcom development…?

Is the QNX target expected to be part of the open sourced Watcom?

For official word on this, you’ll all have to wait for an official
statement from our corporate communications dept. However, we have
in the past officially stated that we support gcc for NTO development,
so that is a known issue.

Also, our official compiler for QNX4 is still Watcom 10.6, and we still
have to officially charge money for it.

When and how this will change is unknown at this point, and any official
word will not likely be disseminated through these newsgroups.

So I’d really rather not have folks building and running on the same box.

Understood. I think what Mario is getting at is that most people develop
on one box that runs QNX or NTO, and target another box running QNX
or NTO.

I don’t doubt that QNX is quite capable, the point is that we already have an
ANSI C source tree, build procedures and a lot of investment in those. I want
to limit the exposure to the QNX system details to a couple of folks who will
know all about it, and keep everyone else blissfully ignorant of what happens
inside the magic box. The application shouldn’t care what’s under the hood as
long as the operating system abstractions look the same to it. (at this point
a grand piano falls from a height of a thousand metres and destroys most of
that ideal world, but that’s the goal anyway)

Understood. You want to cross develop, and let your developers stay on
the systems they already know.

[Rick_Lake1]

Igor Kovalenko wrote:

The bottom line, he’ll have to buy Watcom 10.6 from QSSL anyway, to get
libraries/headers. Then use either GCC or Watcom 11 on Windows to
cross-compile. I don’t know if it is possible to link QNX4 executables
on Windows with Watcom.
[…]

I have Watcom 11 for windows and I actually managed to modify some
config files to build QNX4 executables (using the 10.6 QNX4 libraries)
with the IDE. I haven’t really pursued this option much, though. The
most important reason being because I couldn’t debug executables.


regards,

rick

[Brown_Richard]

I have Watcom 11 for windows and I actually managed to modify some
config files to build QNX4 executables (using the 10.6 QNX4 libraries)
with the IDE. I haven’t really pursued this option much, though. The
most important reason being because I couldn’t debug executables.


regards,

rick

We have Watcom 10.6 for windows. With a few mods to some key config files
and the QNX4 libraries we can compile and debug. The only thing failing for
us is the profiler but I think that is due to C++. We are able to profile
our C source executables fine but C++ fails miserably. I tried to pursue it
a few times but priorities shifted.

We use NT as our host and samba mounted QNX as the target. We can then have
the executables dumped directly to the target for testing, etc. We used to
due all our development on QNX for QNX but found we were limited to a select
set of tools. Moving to NT as the host we now have a little more flexibility
with the development/support tools we choose. Watcom IDE isn’t great but I
find its simpler than what we had before. I don’t mean this as a knock
against QNX. QNX is a great platform for the end system we are building.

We also have Watcom 11.0 but there were some significant changes from 10.6
to 11.0. I think there may be some compatibility issues use the 11.0 toolset
and 10.6 libraries so when the 11.0 Beta was cancelled we de-installed 11.0
and moved back to the proven 10.6.

• Richard