Mailing Lists: March 2022

Wednesday, 30 March 2022

[gccsdk] Libraries

Hi all

As you might've guessed from my earlier messages, I've been updating Python to build with GCC10. I now have Python 3.10.4 built with GCC10, along with the stabdard library extensions that work on RISC OS (which is most but not all of them).

To allow me to do this, I've had to keep a separate tree of development libraries for both GCC10 and GCC4. I've not installed any of the -dev packages I've made via packman as I think the GCC10 versions would clash with the GCC4 ones - same package, different files.

I believe the shared libs used at runtime are okay as they live within a different directory within !SharedLibs, but is there a similar thing that can be done for the dev libs? I guess this is only an issue for native building.

A "not thought about it much" idea is something like this - taking LibBZ2 as an example

LibBZ2$abi2/0$Path would be <LIbBZ2$Dir>,<LibBZ2$Dir>.abi2/0.

LibBZ2$armeabihf$Path would be <LIbBZ2$Dir>,<LibBZ2$Dir>.armeabifh.

LibBZ2$Path need to mirror LibBZ2$abi2/0$Path to keep existing things working.

where <LibBZ2$Dir> contains directories h (for the headers) abi2/0 and armeabihf, each of which has a libbz2/a in it for the abi in question.

I'm sure someone will now point out why it won't work :)

Cheers

Chris

Re: [gccsdk] OpenSSL 1.1.1

Update on this .. a bit of hacking and I've got a working LibOpenSSL 1.1.1 form armeabihf. I had to tell it to use armv4, and remove the compiled binaries (which it got upset about not being for ARM - I assume it meant not form ARMv3)

On 29/03/2022 19:08 Chris Johns <chris@lessthan3.org.uk> wrote:

Digging up an old thread ..

I couldn't get this to build with the cross compiler targeting output to work with gcc10.

The first hurdle was 1.1.1n wasn't available. I tried 1.1.1m instead but it then seemed to suggest it couldn't target armv3, I tried to take that out and it still wasn't happy so I gave up for now.

I think I got most of the other Libs to build.

Cheers

Chris

Sent from my iPhone

On 14 Oct 2021, at 13:12, Chris Gransden <chrisg@care4free.net> wrote:

In article <961CC205-3958-4323-B5D3-E4D00D4EC7FD@lessthan3.org.uk>,

Chris Johns <chris@lessthan3.org.uk> wrote:

Hello

I'm porting python 3.10.0 to risc os, and while it now builds along with most of the extensions, it looks like python now requires libssl 1.1.1.

Does anyone know the state of that for RISC OS? Is it not done because there's been no real news or is there something more significant?

libssl1.1 should appear in PackMan tomorrow. There is also an updated

libssl (1.0.2) so they can both be installed at the same time.

Update current libssl (1.0.2) before installing libssl1.1.

_______________________________________________

GCCSDK mailing list gcc@gccsdk.riscos.info

Bugzilla: http://www.riscos.info/bugzilla/index.cgi

List Info: http://www.riscos.info/mailman/listinfo/gcc

Main Page: http://www.riscos.info/index.php/GCCSDK

_______________________________________________

GCCSDK mailing list gcc@gccsdk.riscos.info

Bugzilla: http://www.riscos.info/bugzilla/index.cgi

List Info: http://www.riscos.info/mailman/listinfo/gcc

Main Page: http://www.riscos.info/index.php/GCCSDK

Tuesday, 29 March 2022

Re: [gccsdk] OpenSSL 1.1.1

Digging up an old thread ..

I couldn't get this to build with the cross compiler targeting output to work with gcc10.

The first hurdle was 1.1.1n wasn't available. I tried 1.1.1m instead but it then seemed to suggest it couldn't target armv3, I tried to take that out and it still wasn't happy so I gave up for now.

I think I got most of the other Libs to build.

Cheers

Chris

Sent from my iPhone

> On 14 Oct 2021, at 13:12, Chris Gransden <chrisg@care4free.net> wrote:
>
> In article <961CC205-3958-4323-B5D3-E4D00D4EC7FD@lessthan3.org.uk>,
> Chris Johns <chris@lessthan3.org.uk> wrote:
>> Hello
>
>> I'm porting python 3.10.0 to risc os, and while it now builds along with most of the extensions, it looks like python now requires libssl 1.1.1.
>
>> Does anyone know the state of that for RISC OS? Is it not done because there's been no real news or is there something more significant?
>
> libssl1.1 should appear in PackMan tomorrow. There is also an updated
> libssl (1.0.2) so they can both be installed at the same time.
>
> Update current libssl (1.0.2) before installing libssl1.1.
>
>
> _______________________________________________
> GCCSDK mailing list gcc@gccsdk.riscos.info
> Bugzilla: http://www.riscos.info/bugzilla/index.cgi
> List Info: http://www.riscos.info/mailman/listinfo/gcc
> Main Page: http://www.riscos.info/index.php/GCCSDK

_______________________________________________
GCCSDK mailing list gcc@gccsdk.riscos.info
Bugzilla: http://www.riscos.info/bugzilla/index.cgi
List Info: http://www.riscos.info/mailman/listinfo/gcc
Main Page: http://www.riscos.info/index.php/GCCSDK

Re: [gccsdk] gcc Digest, Vol 164, Issue 6

Hello all!

I was asked by Norman Lawrence if it is possible to make also available
Fortran to RISC OS as it is also part of the GCC package.

Anybody in the mailing list that is interested or capabale to compile it?

regards

Stefan Fröhling

Am 29.03.2022 um 02:00 schrieb gcc-request@gccsdk.riscos.info:
> Send gcc mailing list submissions to
> gcc@gccsdk.riscos.info
>
> To subscribe or unsubscribe via the World Wide Web, visit
> http://www.riscos.info/mailman/listinfo/gcc
> or, via email, send a message with subject or body 'help' to
> gcc-request@gccsdk.riscos.info
>
> You can reach the person managing the list at
> gcc-owner@gccsdk.riscos.info
>
> When replying, please edit your Subject line so it is more specific
> than "Re: Contents of gcc digest..."
>
>
> Today's Topics:
>
> 1. Re: Using GCC 10 to build python (Chris Johns)
>
>
> ----------------------------------------------------------------------
>
> Message: 1
> Date: Mon, 28 Mar 2022 14:38:50 +0100 (BST)
> From: Chris Johns <chris@lessthan3.org.uk>
> To: Lee Noar <lee.noar@sky.com>, GCC SDK <gcc@gccsdk.riscos.info>
> Subject: Re: [gccsdk] Using GCC 10 to build python
> Message-ID: <362801427.474638.1648474730125@email.ionos.co.uk>
> Content-Type: text/plain; charset="us-ascii"
>
> An HTML attachment was scrubbed...
> URL: <http://www.riscos.info/pipermail/gcc/attachments/20220328/b384da36/attachment-0001.html>
>
> ------------------------------
>
> _______________________________________________
> gcc mailing list
> gcc@gccsdk.riscos.info
> http://www.riscos.info/mailman/listinfo/gcc
>
> End of gcc Digest, Vol 164, Issue 6
> ***********************************

--
Diese E-Mail wurde von AVG auf Viren geprüft.
http://www.avg.com

Monday, 28 March 2022

Re: [gccsdk] Using GCC 10 to build python

I managed to make some progress over the weekend and I've got the python binary built with GCC 10 (natively).

One thing I have found is that closing a taskwindow that's running python takes the machine out. Quiting python first is okay, but if you don't it totally hangs.

Cheers

Chris

On 12/03/2022 10:48 Lee Noar <lee.noar@sky.com> wrote:

On 09/03/2022 14:31, Chris Johns wrote:

Hello

I'd like to see if Python 3.10 will build with GCC 10.

I suspect there are a few issues that might come into play here. One is

that I usually build native, so it might take a week to build (IIRC GCC

10 is very slow when building native). I could probably get it to

cross-compile on my unix box though.

Cross-compiling is the better option because you have the benefit of the

autobuilder to build the dependencies. After building GCC 10, adding

RO_USE_ARMEABIHF=yes

to your build-setvars ensures that most things are then built with GCC

10 (there may be some old packages that make assumptions about the

compiler in use, but I think those are rare).

The second is the various shared libs that python uses.Am I right in

thinking that I would need to get (or build) compatible versions of them

all to work with the output of gcc 10?

Yes, GCC 10 uses a different stack extension strategy than previous

versions. Instead of building stack checks into the code and adding

new chunks to extend the stack, GCC 10 uses a flat stack and allocates

new stack memory pages on the fly via an abort handler. This has the

benefit of an extra work register (i.e. R10) and a stack layout that

matches what ported code is expecting (e.g. JITs, exception handlers).

Is does, however, mean that GCC 4 code and GCC 10 code are not

compatible.

Lee.

Sunday, 27 March 2022

Re: Distoted colours in HTML documents

In message <59cf38bc5aGroverR@orpheusmail.co.uk>
Rod at Orpheusmail <GroverR@orpheusmail.co.uk> wrote:

> For sometime now NetSurf has been exhibiting colour distortion. At least, I
> don't know what else to call it. In pictures, contained in html files for
> example, reds appear blue, as do peoples faces, Pale blue appears yellow
> while dark blue appears brown. And so on...

> I don't know what causes this problem or what to do to correct it. Can
> anybody help please?

At a guess, try opening NetSurf's Choices window from the iconbar menu,
click on the Images icon, and check that both foreground and background
image settings are "Use OS".

Bryan.

--
RISC OS User Group Of London - https://www.rougol.jellybaby.net/
RISC OS London Show - https://www.riscoslondonshow.co.uk/
_______________________________________________
netsurf-users mailing list -- netsurf-users@netsurf-browser.org
To unsubscribe send an email to netsurf-users-leave@netsurf-browser.org

Re: Distoted colours in HTML documents

In article <mpro.r9ebuf004d1np0h75.pittdj@pittdj.co.uk>,
David Pitt <pittdj@pittdj.co.uk> wrote:
> Rod at Orpheusmail, on 26 Mar, wrote:

>> For sometime now NetSurf has been exhibiting colour distortion. At least,
>> I don't know what else to call it. In pictures, contained in html files
>> for example, reds appear blue, as do peoples faces, Pale blue appears
>> yellow while dark blue appears brown. And so on...

[snip]

> In NetSurf's Choices, Images set both Foreground and Background to 'Use OS'.

Sorted! Thank you David.
Thanks also to everyone else who made suggestions.

Kind regards,
Rod Grover

--
Rod Grover
GroverR@orpheusmail.co.uk
_______________________________________________
netsurf-users mailing list -- netsurf-users@netsurf-browser.org
To unsubscribe send an email to netsurf-users-leave@netsurf-browser.org

Re: Distoted colours in HTML documents

In article <59cf38bc5aGroverR@orpheusmail.co.uk>,
Rod at Orpheusmail <GroverR@orpheusmail.co.uk> wrote:
> For sometime now NetSurf has been exhibiting colour distortion. At
> least, I don't know what else to call it. In pictures, contained in
> html files for example, reds appear blue, as do peoples faces, Pale
> blue appears yellow while dark blue appears brown. And so on...

When did this start? Has it worked normally previously on the
Titanium? Remember - some of the later RISC OS hardware had the R and
B channels swapped and there is a new RISC OS 5 spriteword format. It
does sound like the RGB versus BGR problem.

As far as I am aware NetSurf copes with the colour swap.

--
Chris Johnson
Edinburgh
_______________________________________________
netsurf-users mailing list -- netsurf-users@netsurf-browser.org
To unsubscribe send an email to netsurf-users-leave@netsurf-browser.org

Re: Distoted colours in HTML documents

> For sometime now NetSurf has been exhibiting colour distortion. At
> least, I don't know what else to call it. In pictures, contained in
> html files for example, reds appear blue, as do peoples faces, Pale
> blue appears yellow while dark blue appears brown. And so on...

What else has changed? For example, did this start when you switched
to a new version of netsurf, or a new version of your OS, or new
graphics hardware, or any such?

I once saw something very similar with an X program on Linux: all
colours came out with the red and blue primaries switched. In that
case it was a bug in the program - it was assuming that the primaries
were presented in a particular order, rather than paying attention to
the information X provides about how they are actually arranged. (I
assume some developer mistook "my system does it this way" for "this is
how it always works".)

You say RISC OS. I don't know what that uses for graphics interfacing,
but it is not a great stretch for me to imagine something similar
happening there: a new graphics device, a new version of some software
layer, something of the sort. I could also imagine it being a bug in
netsurf itself, probably one like the one I sketched above, assuming
something is always a particular when when it actually isn't.

Without more information (most of which I'm not competent to evaluate,
since I don't know RISC OS and I don't really know much about NetSurf),
it's hard to do more than speculate.

/~\ The ASCII Mouse
\ / Ribbon Campaign
X Against HTML mouse@rodents-montreal.org
/ \ Email! 7D C8 61 52 5D E7 2D 39 4E F1 31 3E E8 B3 27 4B
_______________________________________________
netsurf-users mailing list -- netsurf-users@netsurf-browser.org
To unsubscribe send an email to netsurf-users-leave@netsurf-browser.org

Saturday, 26 March 2022

Distoted colours in HTML documents

For sometime now NetSurf has been exhibiting colour distortion. At least, I
don't know what else to call it. In pictures, contained in html files for
example, reds appear blue, as do peoples faces, Pale blue appears yellow
while dark blue appears brown. And so on...

I don't know what causes this problem or what to do to correct it. Can
anybody help please?

I'm using NetSurf v3.11 (Dev CI#5327), but I get the same effect when using
v3.10
The machine - a Titanium running RISC OS 5.28(19-Oct-20).

I would very much appreciate some help here please.

Kind regardss,
Rod Grover

--
Rod Grover
GroverR@orpheusmail.co.uk
_______________________________________________
netsurf-users mailing list -- netsurf-users@netsurf-browser.org
To unsubscribe send an email to netsurf-users-leave@netsurf-browser.org

Thursday, 17 March 2022

Re: [gccsdk] DeskLib

On 11 Mar, Theo Markettos wrote in message
<20220311235417.4mamo4bgvuchjlxw@chiark.greenend.org.uk>:

> On Fri, Mar 11, 2022 at 09:48:38PM +0000, Cameron Cawley wrote:
>
> > I've noticed that the ccres subversion repository has recently been
> > migrated to GitHub. Would the same thing be possible for DeskLib? I'd be
> > interested in contributing some changes, and it would be easier if I
> > could work with branches and pull requests rather than loose patches.
>
> I've had a quick attempt: https://github.com/riscos-dot-info/DeskLib

That's useful -- thanks!

> If anyone on that list wants to provide an email address they are happy to
> live in the git history, please let me know.

> stevef = stevef <stevef>

info@stevefryatt.org.uk is the address that I usually use for GitHub
submissions, and I'm happy for that to be used in any repos migrated from
riscos.info.

> adam = adam <adam>

I'd have to check back through my emails, but I was in contact with Adam a
couple of years ago over WinEd and DeskLib, so I may have asked him about
this.

> I'm not sure if anyone is looking after Desklib these days (Steve F
> committed something almost 2 years ago) so if you or anyone else wants
> direct commit rights to the github repo, do let me know your github
> username.

I was only making changes when necessary for WinEd, but would continue doing
that if anything is required in the future.

Moving the WinEd repo over has been on my to-do list for a while now, but
I've not really had any time for RISC OS work since last autumn, so it had
stalled along with a lot of other things. I've got a migrated copy here, so
should probably look at pushing it to GitHub at some point.

--
Steve Fryatt - Leeds, England

http://www.stevefryatt.org.uk/

Saturday, 12 March 2022

Re: [gccsdk] DeskLib

On Sat, Mar 12, 2022 at 09:16:39PM +0000, Cameron Cawley wrote:
> According to grep, the desklib and desklib-scl recipes were the only ones
> that used AB_SVN_AOF, and the comments in fetch-program state "AB_SVN_AOF
> support is only temporary", so it made sense to remove it. It isn't
> required, though, and the rest of the patch can be applied without the
> fetch-program changes.

Oh, I see. You're right, I don't think AOF build recipes are much use
to us any more, so makes sense to remove it. I've committed that into
gccsdk SVN (r7692).

Thanks for the patch,
Theo

_______________________________________________
GCCSDK mailing list gcc@gccsdk.riscos.info
Bugzilla: http://www.riscos.info/bugzilla/index.cgi
List Info: http://www.riscos.info/mailman/listinfo/gcc
Main Page: http://www.riscos.info/index.php/GCCSDK

Re: [gccsdk] DeskLib

On Sat, 12 Mar 2022 at 20:46, Theo Markettos <theo@markettos.org.uk> wrote:

On the patch to the autobuilder scripts, I see there are some changes to
fetch-program and setvars-template, mostly removing code. Are those
necessary for your patch, or are they stray changes in your tree? Just
wanted to make sure they're not going to affect other autobuilder packages
that do use them (they look like they're to do with SVN support, which I'm
sure other packages use). If you're just refactoring the existing code and
the outcome is the same then that's fine.

According to grep, the desklib and desklib-scl recipes were the only ones that used AB_SVN_AOF, and the comments in fetch-program state "AB_SVN_AOF support is only temporary", so it made sense to remove it. It isn't required, though, and the rest of the patch can be applied without the fetch-program changes.

Regards

Cameron

Re: [gccsdk] DeskLib

On Sat, Mar 12, 2022 at 02:48:21PM +0000, Cameron Cawley wrote:
> Hi Theo
>
> Thanks for sorting that out. I've managed to successfully cross-compile
> DeskLib via the autobuilder with the attached patch. It currently requires
> some changes in the autobuilder branch in my fork, but once the relevant PR
> has been merged, the recipe can be changed to build from the upstream
> repository instead. I've also tried cloning DeskLib on RISC OS using
> SimpleGit, which seems to work but I've been unable to build natively
> because GCC can't locate the DeskLib headers for some reason.

Thanks for the PR. If I may, I'll keep that open until I've done the SVN
email address cleanup (I should email some more people) and then re-convert
the SVN and (possibly manually) rebase the PR on the new repo. Feel free to
open any more PRs in the meantime.

On the patch to the autobuilder scripts, I see there are some changes to
fetch-program and setvars-template, mostly removing code. Are those
necessary for your patch, or are they stray changes in your tree? Just
wanted to make sure they're not going to affect other autobuilder packages
that do use them (they look like they're to do with SVN support, which I'm
sure other packages use). If you're just refactoring the existing code and
the outcome is the same then that's fine.

Thanks
Theo

_______________________________________________
GCCSDK mailing list gcc@gccsdk.riscos.info
Bugzilla: http://www.riscos.info/bugzilla/index.cgi
List Info: http://www.riscos.info/mailman/listinfo/gcc
Main Page: http://www.riscos.info/index.php/GCCSDK

Re: [gccsdk] DeskLib

Hi Theo

Thanks for sorting that out. I've managed to successfully cross-compile DeskLib via the autobuilder with the attached patch. It currently requires some changes in the autobuilder branch in my fork, but once the relevant PR has been merged, the recipe can be changed to build from the upstream repository instead. I've also tried cloning DeskLib on RISC OS using SimpleGit, which seems to work but I've been unable to build natively because GCC can't locate the DeskLib headers for some reason.

Regards

Cameron

On Fri, 11 Mar 2022 at 23:54, Theo Markettos <theo@markettos.org.uk> wrote:

On Fri, Mar 11, 2022 at 09:48:38PM +0000, Cameron Cawley wrote:
> Hi
>
> I've noticed that the ccres subversion repository has recently been
> migrated to GitHub. Would the same thing be possible for DeskLib? I'd be
> interested in contributing some changes, and it would be easier if I could
> work with branches and pull requests rather than loose patches.

I've had a quick attempt:
https://github.com/riscos-dot-info/DeskLib

This is marked as experimental for now - please give it a spin and see if
there's any issues. If there are, I may have to delete it and re-convert.
(if this happens you can likely replay any commits against the re-upload)

In particular, the svn->git userid translation is missing a number of email
addresses (plus myself, John Tytgat and Alan Buckley which I have details
for):

rosery = rosery <rosery>
no-author = no-author <no-author>
alex = alex <alex>
lee = lee <lee>
nick = nick <nick>
admin = admin <admin>
cpalmer = cpalmer <cpalmer>
jjvdgeer = jjvdgeer <jjvdgeer>
cgrasden = cgrasden <cgrasden>
jmb = jmb <jmb>
peter = peter <peter>
cpmartin = cpmartin <cpmartin>
phlamethrower = phlamethrower <phlamethrower>
stevef = stevef <stevef>
adam = adam <adam>

If anyone on that list wants to provide an email address they are happy to
live in the git history, please let me know.

I'm not sure if anyone is looking after Desklib these days (Steve F
committed something almost 2 years ago) so if you or anyone else wants
direct commit rights to the github repo, do let me know your github username.

Theo

Re: [gccsdk] Using GCC 10 to build python

On 09/03/2022 14:31, Chris Johns wrote:
> Hello
>
> I'd like to see if Python 3.10 will build with GCC 10.
>
> I suspect there are a few issues that might come into play here. One is
> that I usually build native, so it might take a week to build (IIRC GCC
> 10 is very slow when building native). I could probably get it to
> cross-compile on my unix box though.

Cross-compiling is the better option because you have the benefit of the
autobuilder to build the dependencies. After building GCC 10, adding

RO_USE_ARMEABIHF=yes

to your build-setvars ensures that most things are then built with GCC
10 (there may be some old packages that make assumptions about the
compiler in use, but I think those are rare).

> The second is the various shared libs that python uses.Am I right in
> thinking that I would need to get (or build) compatible versions of them
> all to work with the output of gcc 10?

Yes, GCC 10 uses a different stack extension strategy than previous
versions. Instead of building stack checks into the code and adding
new chunks to extend the stack, GCC 10 uses a flat stack and allocates
new stack memory pages on the fly via an abort handler. This has the
benefit of an extra work register (i.e. R10) and a stack layout that
matches what ported code is expecting (e.g. JITs, exception handlers).
Is does, however, mean that GCC 4 code and GCC 10 code are not
compatible.

Lee.

_______________________________________________
GCCSDK mailing list gcc@gccsdk.riscos.info
Bugzilla: http://www.riscos.info/bugzilla/index.cgi
List Info: http://www.riscos.info/mailman/listinfo/gcc
Main Page: http://www.riscos.info/index.php/GCCSDK

Friday, 11 March 2022

Re: [gccsdk] DeskLib

On Fri, 11 Mar 2022, Theo Markettos wrote:

> In particular, the svn->git userid translation is missing a number of email
> addresses (plus myself, John Tytgat and Alan Buckley which I have details
> for):
>
> [...]
> phlamethrower = phlamethrower <phlamethrower>
>
> If anyone on that list wants to provide an email address they are happy to
> live in the git history, please let me know.

Feel free to use me@phlamethrower.co.uk for my contributions, across any
of the projects you're converting.

Cheers,

- Jeffrey

_______________________________________________
GCCSDK mailing list gcc@gccsdk.riscos.info
Bugzilla: http://www.riscos.info/bugzilla/index.cgi
List Info: http://www.riscos.info/mailman/listinfo/gcc
Main Page: http://www.riscos.info/index.php/GCCSDK

Re: [gccsdk] DeskLib

On Fri, Mar 11, 2022 at 09:48:38PM +0000, Cameron Cawley wrote:
> Hi
>
> I've noticed that the ccres subversion repository has recently been
> migrated to GitHub. Would the same thing be possible for DeskLib? I'd be
> interested in contributing some changes, and it would be easier if I could
> work with branches and pull requests rather than loose patches.

I've had a quick attempt:
https://github.com/riscos-dot-info/DeskLib

This is marked as experimental for now - please give it a spin and see if
there's any issues. If there are, I may have to delete it and re-convert.
(if this happens you can likely replay any commits against the re-upload)

In particular, the svn->git userid translation is missing a number of email
addresses (plus myself, John Tytgat and Alan Buckley which I have details
for):

rosery = rosery <rosery>
no-author = no-author <no-author>
alex = alex <alex>
lee = lee <lee>
nick = nick <nick>
admin = admin <admin>
cpalmer = cpalmer <cpalmer>
jjvdgeer = jjvdgeer <jjvdgeer>
cgrasden = cgrasden <cgrasden>
jmb = jmb <jmb>
peter = peter <peter>
cpmartin = cpmartin <cpmartin>
phlamethrower = phlamethrower <phlamethrower>
stevef = stevef <stevef>
adam = adam <adam>

If anyone on that list wants to provide an email address they are happy to
live in the git history, please let me know.

I'm not sure if anyone is looking after Desklib these days (Steve F
committed something almost 2 years ago) so if you or anyone else wants
direct commit rights to the github repo, do let me know your github username.

Theo

[gccsdk] DeskLib

I've noticed that the ccres subversion repository has recently been migrated to GitHub. Would the same thing be possible for DeskLib? I'd be interested in contributing some changes, and it would be easier if I could work with branches and pull requests rather than loose patches.

Regards

Cameron

Wednesday, 9 March 2022

[gccsdk] Using GCC 10 to build python

Hello

I'd like to see if Python 3.10 will build with GCC 10.

I suspect there are a few issues that might come into play here. One is
that I usually build native, so it might take a week to build (IIRC GCC
10 is very slow when building native). I could probably get it to
cross-compile on my unix box though.

The second is the various shared libs that python uses.Am I right in
thinking that I would need to get (or build) compatible versions of them
all to work with the output of gcc 10?

Cheers

Chris

_______________________________________________
GCCSDK mailing list gcc@gccsdk.riscos.info
Bugzilla: http://www.riscos.info/bugzilla/index.cgi
List Info: http://www.riscos.info/mailman/listinfo/gcc
Main Page: http://www.riscos.info/index.php/GCCSDK

Tuesday, 8 March 2022

Re: [gccsdk] PDF files

[Posted and mailed]

> Try Mu View, works nice on ARMX6 https://www.riscosports.co.uk/vfp/

A word of warning.

The version of MU from the above link - 0.04 alpha (14.10.2020)
works nicely.

However the later version 0.05 alpha (27.02.2022) from PackMan loops
endlessly on my ARMX6 RISC OS version 5.29 (02 Nov 2020)

John

--
John Rickman

_______________________________________________
GCCSDK mailing list gcc@gccsdk.riscos.info
Bugzilla: http://www.riscos.info/bugzilla/index.cgi
List Info: http://www.riscos.info/mailman/listinfo/gcc
Main Page: http://www.riscos.info/index.php/GCCSDK