[passportowner]

So I made a script to automate the build process, it does everything automatically, but fails with the same error as I told before.

You can find it here:
https://github.com/extrowerk/bb10-toolchain

Have fun, talk less about whatifs and compile more instead.

[BronzeBeard]

Curious why you would want build the latest and greatest version of gcc.

Planning on porting C++21 stuff to the platform? I'm not seeing much of a use case for putting effort bringing the latest compilers to the platform when there are glaring deficiencies in use-ablity, such as native browser alternatives.

Any code you could compile and run on the platform will work with the existing c++11 standards the current gcc supports.

So just curious on the reasoning

My take anyway, good luck on your progress! And hopefully the FUD crowd doesn't derail this thread again.

Posted via CB10

[passportowner]

Because the host system have 10.0.2 and somebody on the net said for similar problem that the gcc and the binutils version mismatch can be the culprit, so I tried to test hypothesis. Sadly there is no GCC7.1 anymore in the repos, so I can't downgrade my host GCC.
My wish is to build GCC7.1 so we can have a modern toolchain so we can update the Qt port to the latest one. Idk if I am capable to do that alone.
We could try to build it on the device, and if it is working we can tell for sure it is actually possible or not.

[passportowner]

I mistakenly assumed the "release_710" means GCC 7.1.0, but it is actually GCC 8.3.0.

Probably the 710 means the QNX SDK version, sorry for the misunderstanding.

[passportowner]

I got this:

[szilard@Boxy bb10-toolchain]$ /opt/qnx800/host/linux/x86_64/usr/bin/arm-unknown-nto-qnx8.0.0eabi-gcc-8.3.0 -v
Using built-in specs.
COLLECT_GCC=/opt/qnx800/host/linux/x86_64/usr/bin/arm-unknown-nto-qnx8.0.0eabi-gcc-8.3.0
COLLECT_LTO_WRAPPER=/opt/qnx800/host/linux/x86_64/usr/bin/../lib/gcc/arm-unknown-nto-qnx8.0.0eabi/8.3.0/lto-wrapper
Target: arm-unknown-nto-qnx8.0.0eabi
Configured with: ../bb10-gcc/configure --srcdir=../bb10-gcc --build=x86_64-unknown-linux-gnu --enable-cheaders=c --with-as=arm-unknown-nto-qnx8.0.0eabi-as --with-ld=arm-unknown-nto-qnx8.0.0eabi-ld --with-sysroot=/opt/qnx800/target/qnx6/ --disable-werror --libdir=/opt/qnx800/host/linux/x86_64/usr/lib --libexecdir=/opt/qnx800/host/linux/x86_64/usr/lib --target=arm-unknown-nto-qnx8.0.0eabi --prefix=/opt/qnx800/host/linux/x86_64/usr --exec-prefix=/opt/qnx800/host/linux/x86_64/usr --with-local-prefix=/opt/qnx800/host/linux/x86_64/usr --enable-languages=c++ --enable-threads=posix --disable-nls --disable-tls --disable-libssp --disable-libstdcxx-pch --enable-libmudflap --enable-libgomp --enable-__cxa_atexit --with-gxx-include-dir=/opt/qnx800/target/qnx6/usr/include/c++/8.3.0 --enable-shared --enable-multilib --with-bugurl=http://www.qnx.com --enable-gnu-indirect-function --enable-stack-protector --with-float=softfp --with-arch=armv7-a --with-fpu=vfpv3-d16 --with-mode=thumb CC=gcc-8 LDFLAGS='-Wl,-s ' AUTOMAKE=: AUTOCONF=: AUTOHEADER=: AUTORECONF=: ACLOCAL=:
Thread model: posix
gcc version 8.3.0 20190222 (EXPERIMENTAL) QNX 710 (GCC)

[m3mb3rsh1p]

Too much "What if" talk here lately, let's return to more interesting waters.

So I have successfully built binutils, mpfr, mpc, gomp from the foundry27's "release710" sources (and the latest vanilla versions from gnu) on Linux targeting our mobile devices.

...

Also you definitely not want to build GCC on the phone.

Best Regards.

Yay! THANK YOU! I will definitely look at this. I'm working on some BB10 stuff right now so a more current gcc will be useful.

btw, I love that you feel the need to advise against building gcc on the phone... That's what I love about BB10; so many possibilities.

[m3mb3rsh1p]

Curious why you would want build the latest and greatest version of gcc.

Even though the native browser is outdated, there are other Apps that BB10 users would appreciate. Productivity apps, games, system utilities are still useful. Most C++ apps and libraries are now written with the C++14 standard as the minimum, with C++17 being the default target so a compiler update would avail us of a wider software selection.

[passportowner]

We just tested GCC with a simple hello world, it builds, links and even better: it works on the phone!

[passportowner]

I just noticed BB Inc updated the GCC repository in Foundry27 some days ago, so we need to update our port, nobody want to run some old program on his phone, we need to stay up to date!
Also we need to get run it on the phone too, because I hate linux.

That's the plan for morning.

[passportowner]

Also where can I find the BB Qt5.6 sources?

[m3mb3rsh1p]

I'm having a bit of trouble finishing the gcc compilation (probably because I don't use /opt) but I'll figure it out.
Qt includes QNX support with the standard source package (I have the latest sources and the QNX screen plugin is there). I don't know what is needed to get widgets working on BB10, though.

[passportowner]

Try to run the commands one by one in a terminal and don't use parallel builds so you will see where it fails, probably some build time dependency is missing. Or just download the prebuilt x86-64 package from github.

[Leyra B10]

Even though the native browser is outdated, there are other Apps that BB10 users would appreciate. Productivity apps, games, system utilities are still useful. Most C++ apps and libraries are now written with the C++14 standard as the minimum, with C++17 being the default target so a compiler update would avail us of a wider software selection.

It would be cool to include bindings for web works api's and cascades. So dot notation object code library exposing a global api to map a useable grammar like CRUD to. Or set and get for headers if you will.

Up to date tls extension too.

Posted via CB10

[passportowner]

There are problems with libstdc++ which I haven't noticed yesterday. While g++ was installed the supporting lib not because problems with the math headers. I assume the sys headers in the NDK are just too old.
This kind of problems are way beyond my skillset to fix, I am a mechanical engineer, not a programmer. So as an alternative I went to the next recent BB GCC release, it is 5.4.0, which I was able to compile fully.
While it is not as fancy as GCC8, this is still a relatively modern compiler so we can use it to build Qt5 with it (we used it for long on Haiku).

Sorry for the misunderstanding.

Btw the C compiler part of GCC8 works if you want to build C stuff, sadly Qt is C++.

[Leyra B10]

Btw the C compiler part of GCC8 works if you want to build C stuff, sadly Qt is C++.

I tried BerryFarms fork of mordaks cli tools and term48. That works as well.

Posted via CB10

[anon(5597702)]

I tried BerryFarms fork of mordaks cli tools and term48. That works as well.

Posted via CB10

I really liked term48 when I had a use for it.

Posted via CB10

[BronzeBeard]

Also where can I find the BB Qt5.6 sources?

I haven't looked at em yet, but the fixes for qtweb 5.6 are here:
https://github.com/sgear256/qtwebkit/tree/5.6-bb

5.6 can be gotten from QT's website. I linked the build instructions around page 5-10. (The thread gets derailed around page 12).

Posted via CB10

[DonHB]

That is/was BlackBerry's mistake. I will relate my own personal experience. I worked for a company where we developed some specialized handheld devices. It was around the same time Android was available (before there were even phones).

When it came down to evaluating potential OS, QNX did not make the cut. The reason is simply there were so many free options and Android was free to download, modify and try. (Yes, I did manage to get it compiled and running on our hardware).

Ongoing or even onetime license fees are not a big issue. It's the barriers to trying/evaluating it and the software development tools & libraries.

That is probably the single biggest reason Android 'won'. As an engineer I could try it for free and if it was decent, pitch it to the management as an option.

To even try QNX? So many hurdles, and probably need to get lawyers involved with NDA's and such. You have to really, really want it for some reason.
BlackBerry may claim they don't have resources to support small customers but when I was dealing with Android, I never had support from Google. Engineers will figure it out, especially with the help of community forums.

Posted via CB10

But Android was an open-source project, as is Linux.

How would it had benefitted QNX (pre, or post BlackBerry) to give itself up for free?

Even prior to the acquisition, they had 250 employees, revenues of $30-50 million, and a valuation (based on the purchase price) of $200 million.

Sorry to those getting back to work, but I wish to respond to this.

The point Efat made is that the difficulty of acquiring a license not the ultimate license cost dissuades many developers from choosing QNX. That developers want to "kick the tires" prior to committing to a platform and that they are ready to invest development time to experiment. What they are not ready to do is jump through hoops to be able to experiment which is what BlackBerry makes them do. This is where OSS and Linux in particular wins.

My suggestion of a free developers'/non-commercial license is precisely intended to address the experience that EFat had. The idea here is for developers to freely experiment and non-commercially distribute their work if they so decide to do so. Efats difficulty of acquiring a license is the problem that the idea of providing a free developers'/non-commercial license for the Raspberry Pi 4 Model B addresses.

Efat also describes how he did not need support when experimenting with Android which also aligns with by suggestion that all that is needed in addition to the BSP are tools and docs on-line. That the company already has a BSP for the Raspberry Pi and the associated tools minimizes the investment that QNX needs to make. Further, that the Raspberry Pi Foundation is a non-profit intending to facilitate the learning of technology, BlackBerry has options available for how to present this decision to shareholders.

[conite]

Sorry to those getting back to work, but I wish to respond to this.

The point Efat made is that the difficulty of acquiring a license not the ultimate license cost dissuades many developers from choosing QNX. That developers want to "kick the tires" prior to committing to a platform and that they are ready to invest development time to experiment. What they are not ready to do is jump through hoops to be able to experiment which is what BlackBerry makes them do. This is where OSS and Linux in particular wins.

My suggestion of a free developers'/non-commercial license is precisely intended to address the experience that EFat had. The idea here is for developers to freely experiment and non-commercially distribute their work if they so decide to do so. Efats difficulty of acquiring a license is the problem that the idea of providing a free developers'/non-commercial license for the Raspberry Pi 4 Model B addresses.

Efat also describes how he did not need support when experimenting with Android which also aligns with by suggestion that all that is needed in addition to the BSP are tools and docs on-line. That the company already has a BSP for the Raspberry Pi and the associated tools minimizes the investment that QNX needs to make. Further, that the Raspberry Pi Foundation is a non-profit intending to facilitate the learning of technology, BlackBerry has options available for how to present this decision to shareholders.

But you haven't demonstrated, with numbers, how that model would increase net revenues.

Otherwise you'd have to assume BlackBerry ran them, and decided it was a non-starter.

[Dunt Dunt Dunt]

But you haven't demonstrated, with numbers, how that model would increase net revenues.

Otherwise you'd have to assume BlackBerry ran them, and decided it was a non-starter.

I bet Dan Dodge didn't like the changes that BlackBerry made and argued against them. Both 10 years ago when BlackBerry bought QNX and then again five years ago when he moved on.

Either way I don't see the change now doing what he wants.... Just like I doubt that disk with the QNX desktop was anything more than a "let me try this, cool... ok going to reload Linux now".

[DonHB]

But you haven't demonstrated, with numbers, how that model would increase net revenues.

Otherwise you'd have to assume BlackBerry ran them, and decided it was a non-starter.

It is interesting that you don't even acknowledge that Efat's experience as even an issue that should be addressed. Also, you don't acknowledge that the product that is intended to be free is already a product. Further, you don't acknowledge that because little additional investment is needed the primary question is what is the risk vs reward rather than predominantly RoI.

Lastly, do have the inside information to produce those numbers you keep asking for? One information point is: have potential or existing customers have expressed concern regarding availability of developers experienced in QNX. This proposal would address this. Have existing customers been dissuaded from perusing a 30-day trial because it is too short making its use a waste of time. This proposal would address this too.

How would you suggest calculating increased revenue from just addressing the described issues. This is more a judgment call that weighs the risks against the rewards. Like you, BlackBerry may not even acknowledge that these are problems that needs to be addressed. You just repeat the same rote remarks feigning discussion.

[conite]

It is interesting that you don't even acknowledge that Efat's experience as even an issue that should be addressed. Also, you don't acknowledge that the product that is intended to be free is already a product. Further, you don't acknowledge that because little additional investment is needed the primary question is what is the risk vs reward rather than predominantly RoI.

Lastly, do have the inside information to produce those numbers you keep asking for? One information point is: have potential or existing customers have expressed concern regarding availability of developers experienced in QNX. This proposal would address this. Have existing customers been dissuaded from perusing a 30-day trial because it is too short making its use a waste of time. This proposal would address this too.

How would you suggest calculating increased revenue from just addressing the described issues. This is more a judgment call that weighs the risks against the rewards. Like you, BlackBerry may not even acknowledge that these are problems that needs to be addressed. You just repeat the same rote remarks feigning discussion.

Keep dreaming if it makes you feel better. BlackBerry has already made its decision.

[Leyra B10]

Your license idea isn't bad at all. Maybe no one is interested, but either party has an opportunity to share knowledge and accumulate statistics.

[passportowner]

The whole Crackberry is just "BB should do this and that, why can't they see???" rant.

Also this thread is about breathing new life into BB10, but:
- BB have expressed they aren't interested in BB10 anymore
- RPI have nothing to do anything with BB10
- QNX is not BB10 and BB10 is not QNX (they are 2 distinct products, even if BB10 uses QNX kernel
- RPI, QNX, marketing, licensing won't do anything favour for BB10

I think you guys should accept you are kind of last mohikans (which isn't anything special at all), or do what you can. And this topic is just about that: what we can do to keep the ecosystem usable for some more time, talking about system hacks or tricks, etc.

But if you can only talk about unrelated things, you can do it in many different threads, why do you feel entitled yourself to bring unrelated topics into a specific thread?
This is impolite and because this behaviour the signal:noise ratio is pretty low. Just look at the page count. Maybe you guys should look at yourselves from an outer viewpoint.

Whatever, you will probably continue the argumentation till crackberry goes down and even after in unrelated forums.

Also old people says: we got 2 ears and one mouth to listen 2 times more than we talk... Maybe that's the only mistake what BB did: they gave us a keyboard to write to much bs.

[DonHB]

It was in part to keep the thread alive (nothing new has occurred in months), but perhaps you or the OP would create a new thread that summarizes the state of affairs with links to code. Once this is done a moderator can close this thread and provide a link to the new thread.