Show Posts

This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.


Messages - Dave Yeo

Pages: [1] 2 3 ... 382
1
Programming / Re: cross-os2emx
« on: Today at 03:03:57 am »
Yes, it is easy enough not to use lxlite, just some makefile editing, and there's omfstrip or whatever it is called. Same with other things like REXX scripts, which might need porting to sh scripts or install Regina Rexx, which would still need minor makefile editing. Rexx.cmd becomes rexx.rx.
Need to differentiate between native and cross compiling some how to have some if else logic.

2
Programming / Re: cross-os2emx
« on: March 15, 2026, 11:48:08 pm »
Hi David. As far as I can see, it is going to depend on whether OS/2 specific features are needed. eg, Python or Perl would generally work but our ports have OS/2 stuff that a Linux build wouldn't have. Generally they could be added.
Don't know node.js but it might just work. I'm planning on trying to build Qt5 on Linux and see how far it gets.
Really, eventually we need an OS/2 node.js anyways as a bunch of stuff uses it.
There will be other OS/2 stuff that needs rebuilding on Linux, tried FFmpeg and the stock build died from no lxlite.
Rexx is another consideration, thinking of trying to build FF for a test, there's a REXX script where the build would have to be adjusted to use the sh version and I think a couple of batch files that might need converting. It also depends on Lxlite during the packaging step. Need to see if there's a Linux Pascal that will build a Linux lxlite.

3
Applications / Re: Java applications
« on: March 15, 2026, 11:38:32 pm »
Only a white screen here too. Looking at the README.OS2, it looks pretty hit and miss and may need an older Java. Colour depth might matter too, with it wanting 64K colours.
Seems I might have had it running years back but not sure.
https://github.com/OS2World/JAVA-APPS-WeirdX/blob/master/README.OS2

4
Internet / Re: Mypal68: A possible alternative browser?
« on: March 15, 2026, 08:58:53 am »
they could have found a skilled developer to update our existing browser years ago, before we fell so far behind.

Did Ecomstation contribute to the browser? Or something changed that they didn't need to?

Yes, first they paid Rich to fix the printing. That was FF v4. Walter and I took it to v10, then Serenity paid Bitwise for 17 to 45, actually at the end the community was paying. I supported SM and TB, easy once Firefox built and ran as they all use the same engine. At that point I was left alone with a few resources from Arca Noae and I don't have the skills to do what was needed for 52 and it took awhile for me to even get it building to the point where we needed someone much more knowledgeable. I also ported it to the newer GCC and applied a lot of security fixes from 10four (Mac fork of 45).
In theory someone knowledgeable  could continue development to the latest PaleMoon with our source, though PaleMoon has really rearranged the source.
Meanwhile Bitwise thought that porting Chromium would be a better route using Qt. Turned out pretty hard, newer GCC was needed, lots of libc updates and got Qt5 to where it is now. Development kind of stopped due to issues updating Qt5 on our system and then war broke out and the developer left Russia and spent quite a bit of time stateless. Paul applied the Qt5 patches to Qt6 but wasn't skilled enough to really go on. One problem is that Qt6 doesn't officially support 32 bits, so no testing by the Qt folks and no worries about freeing memory at times.
At the beginning of the month, Dmitriy, Bitwise's lead developer, did post that he expects to get back to working on Qt6 in the next couple of months.
Meanwhile KOMH has ported our build system to 64bit Linux which should really help. Memory has been a problem compiling browsers since FF10 where I needed VAL=3072 and at the time with only 1.5 GB of actual ram, the swap file could overflow (2.1GB limit) and crash the system if I did anything else while building. Linking some of the Chromium DLL's is even harder due to the memory limits of our system.
It's actually amazing what we've accomplished in the way of compiling browsers. We were the last 32 bit OS that could build Firefox and considering that most of the design of OS/2 was done in the early 90's or earlier...


5
Internet / Re: Mypal68: A possible alternative browser?
« on: March 14, 2026, 05:56:51 am »

I'm not sure, but I thought that an existing Winflector 32-bit linux client would mean a fairly easy port to OS/2.

Not really, unless it uses Qt which is about the only widget set we share with Linux. The Windows 32 bit version would actually be easier as Windows and OS/2 are cousins. It would still be hard as hell, perhaps easier to go the Odin route. A small team might be able to do it in a year including testing etc to make it stable. Can't see it making business sense.

Quote
So, the bigger problem is that they think OS/2 is too "niche". Unless AN gets involved, it may be hard to dissuade them from this view ... if there are no seat numbers, there's no way to judge interest.

How do we get AN, EComStation, and OS/2 Warp seat count? I can see potential seat numbers in the thread view counts on OS2World ... other sources?

It's hard to say how many OS/2 users there still are, but I doubt there's many using OS/2, eCS and AOS full time. Even if we knew how many licenses Arca Noae has sold, it wouldn't mean much as a lot of them are likely tyre kickers. I've seen a few people from the past show up with an AOS license, play a bit and move on.
AN itself doesn't have much in the way of resources, otherwise they could have found a skilled developer to update our existing browser years ago, before we fell so far behind.

6
Web applications / Re: Need memcached module for php7!
« on: March 13, 2026, 09:56:39 pm »
Does it depend on having PHP installed? I'd guess so

7
Utilities / Re: SETTIME - which time server still supports port-13?
« on: March 13, 2026, 08:13:46 pm »
Well, there is the one that ships with ArcaOS, "The Clock Synchronization Utility" somewhere on your ISO. For command line, os2_ntpd, https://hobbesarchive.com/Home/Download?path=/Hobbes/pub/os2/util/time/NTPD_1-7b.zip

8
Internet / Re: DASH (Dooble Awesome Shell)
« on: March 12, 2026, 04:08:42 am »
It depends on the computer but personally I think the stock VAL value is way to low. You should be able to raise it until it hits PCI space. Here, my computer is stable with 2944 and crashy with 3072. 2560 should be good for most.

9
Internet / Re: DASH (Dooble Awesome Shell)
« on: March 11, 2026, 08:37:05 pm »
I use it quite a bit, though usually use SM. There are tricks. The latest ACPI helps. Installing the block lists etc to stop ads helps and staying away from some multimedia helps. Need a higher VIRTUAADDRESSLIMIT as well. I do find the Qt5 version more stable, while the Qt6 version handles more web sites. Really depends on the sites you use.

10
Internet / Re: DASH (Dooble Awesome Shell)
« on: March 11, 2026, 04:42:46 pm »
Yes, it is more of a curiosity then very useful, especially since copy and paste don't seem to work. Still interesting what can be done and shows a weakness in our Qt6 port.

11
Internet / DASH (Dooble Awesome Shell)
« on: March 11, 2026, 02:21:26 am »
So the latest release of Dooble introduces a new thing, to quote,
Code: [Select]
Introduced the Dooble Awesome Shell (DASH). Interactive commands are not functional. Unix onlySo I had to try it.
With Dooble-Qt5, it works well. You get a bash shell and you can even enter cmd.exe and get a cmd shell. Seems to work as expected. There are limitations such as no interactive commands and starting a PM app blocks the shell until it exits. In cmd.exe can use start to work around it. The bash equivalent doesn't work. 4OS2 is also somewhat broken due to trying to use colours in the prompt and such.
Unluckily with Dooble-Qt6, attempting to run any command results in the tab closing. Perhaps other versions of Qt6 work better. It's a shame as I was thinking of permanently enabling it.
Binaries for those that want to test,
ftp://dry@ftp.os2voice.org/tmp/dooble.2026.03.09-Qt5.zip
ftp://dry@ftp.os2voice.org/tmp/dooble.2026.03.09-Qt5.zip 

12
The eCS system will work. You have to use the proper tools to set up the stick. On the ISO is the directory AOSBOOT containing aosboot.cmd, aosboot.text and the WORK directory. The version of AOSBOOT has to match your ISO.
Code: [Select]
D:\AOSBOOT>aosboot.cmd

  usage: aosboot <drive:> <path to .ISO file>

  example: aosboot Q: C:\downloads\ArcaOS-5.1.iso

  This script creates a bootable USB drive you can use
  to install ArcaOS. It supports booting both BIOS and
  UEFI-based systems. Any USB drive >= 2gb can be used
  as-is.  If the disk has more than one partition, you
  can enter any of the drive letters assigned to it.

  Warning! This process will repartition the USB drive
           and will DESTROY all of its current content.

  Since copying files to flash drives can be very slow
  this process may take 10-30 minutes.

  For additional details, please read "aosboot.txt"

  AosBoot v2 (c)2017,2019 RL Walsh, (c)2017,2019 Arca Noae

I can't remember if eCS came with a easy way to mount the ISO, on AOS oo.exe will do it, oo path\to\ISO. eCS might have had it.

13

I have the ArcaOS-5.1.1-en.iso file copied to the USB drive. I have tried connecting it to each and every USB port on the front and the back, powering off the computer and powering it back on and ... it never sees the USB drive OR it doesn't recognize the file as an ISO file therefore it is not trying to install the contents of the file. Anything else I might be doing wrong?

Sounds weird. Did you correctly set up the stick? OS/2 can't boot of an USB stick, it actually creates a ram disk, copies itself there and boots of the ram disk.
Your first partition on the stick should be similar to this,
Code: [Select]
The volume label in drive I is EFIBOOT.
The Volume Serial Number is 1909:0E31.
Directory of I:\

12-28-25  4:45p        16,047      0 a---  config.sys
12-28-25  4:45p     4,592,762      0 a---  boot1.bin
12-28-25  4:45p    12,213,555      0 a---  boot2.bin
12-28-25  4:45p       402,117      0 a---  boot3.bin
 3-01-26  5:32p         <DIR>      0 ----  efi
        5 file(s)  17,224,481 bytes used
                     55,274,496 bytes free

The efi partition is important along with the boot directory under it and boot containing,
Code: [Select]
8-28-25  9:58a       107,936      0 a---  bootx64.efi
12-28-25  4:45p       785,521      0 a---  os2ldr.bin
 9-01-24 12:18p            13      0 a---  os2ldr.cfg
 9-01-24 12:18p           831      0 a---  ANdb.cer
 9-01-24 12:18p         1,200      0 a---  ANdb.crt

Your system should try to boot bootx64.efi

Quote
By the way, this message board takes two to three minutes to update for me. Is it only me or is there something going on?

Bot attacks and attempts to stop them. It's a crappy situation, the site used to work much better. Martin is working on it.

14
I downloaded the BIOS documentation. Go to the Advanced Menu-->Boot Options. Make sure fast boot is turned off and USB Storage boot is turned on and it should be able to boot an USB stick. It is possible the Boot Order menu will only show USB if a USB device is plugged in.
So you have to get your AOS USB boot stick built and then try again. Might work.
I also don't see any sign of a CSM so you will have to do an UEFI install.

15
Internet / Re: Dooble releases-Qt6
« on: March 08, 2026, 06:31:54 pm »
Actually the problem is the directory separator after the drive letter. set DOOBLE_HOME=h:/tmp\.dooble\ works fine along with various similar settings, while set DOOBLE_HOME=h:\tmp\.dooble\  fails. This affects both Qt5 and Qt6 versions.
Edit: Missed your last edit :)

Pages: [1] 2 3 ... 382