Hyperion Entertainment Message Boards

Hi all,

I am using an A1XE with AmigaOS 4.1 Final Edition with most recent updates; and when I try to use the Turkish locale, I encounter several errors during system boot.

These errors start appearing immediately after enabling the Turkish locale. I don't get any errors with English Locale.

Since the fonts pre-installed with the system do not support Turkish characters, I downloaded and installed fonts with ISO 8859-9 support using System/TypeManager. After installing the fonts, I ran System/FixFonts.

In Prefs/Fonts, I had to set all fonts to the ISO8859-9 compatible ones, otherwise Turkish characters were displayed incorrectly.

⸻

Error 1:

MAKELINK: Can't find "SOBJ:libz.so.1.2.3".

I noticed that my system (updated to AmigaOS 4.1 FE Update 3) includes libz.so.1.2.13.
To resolve this, I manually created a link named libz.so.1.2.3 using the makelink command. I couldn't find which process calls this library with that particular version so I created this workaround for myself.

⸻

Error 2:

CHARSETCONVERT: Unknown FROMCHARSET "UTF-8".
C:Charsetconvert failed, return code 20

These errors appear to be related to notifications. I resolved them by going to:

SYS:Prefs/Notifications → Preferences tab

and selecting one of the ISO8859-9 compatible fonts I installed as the Theme Font.

⸻

Error 3:

MOUNT: "a" is not a valid keyword (file "*.info", line 1, column 1)

During system boot, the system attempts to mount files under DEVS:DosDrivers separately from their corresponding .info icons. Since .info files are not valid mountlists, this results in errors.

To work around this, I modified the S:Startup-Sequence file and added explicit checks for each device, for example:

IF EXISTS DEVS:DOSDRIVERS/PIPE
MOUNT QUIET DEVS:DOSDRIVERS/PIPE
ENDIF

This prevents the system from trying to mount .info files as mountlists.

⸻

Error 4:

MOUNT: "Device" is not a valid keyword
MOUNT: "StackSize" is not a valid keyword

To resolve this, I edited the mountlist files using Notepad and replaced lowercase “i” characters with uppercase “I”.

In Turkish, characters such as I, İ, ı, i have special casing rules, and without proper system-level support, this can lead to parsing issues.

For the same reason, I also had to replace all lowercase “i” characters with uppercase “I” in the following files:
• S:Startup-Sequence
• S:User-Startup

⸻

Additional Observation

When enabling the Turkish language, I noticed that many translations are incomplete or missing.

⸻

Also AmiSSL most recent version to the date I'm writing this message, install script has issues and can not proceed but as soon as I revert back to English, no problem...


Since I have only recently set up the system, I am not yet sure if there are additional issues.

I would greatly appreciate it if these topics could be reviewed in a future update. (Also UTF-8 support.)
You can likely reproduce this behavior easily in your own test environment, but if you need further information or would like me to test anything, I would be more than happy to assist.

rumbler wrote: Wed Mar 25, 2026 2:29 amIn Prefs/Fonts, I had to set all fonts to the ISO8859-9 compatible ones, otherwise Turkish characters were displayed incorrectly.

Don't forget to also change the fonts used by things like NotePad, Shell, AmiDock, MUI, DOpus4 (if you use it), and in the GUI prefs on the Gadgets page (group titles plus the three at the bottom), if they are set.

rumbler wrote: Wed Mar 25, 2026 2:29 am MAKELINK: Can't find "SOBJ:libz.so.1.2.3".

When does this happen, at installation of an update or during normal start?

rumbler wrote: Wed Mar 25, 2026 2:29 am I noticed that my system (updated to AmigaOS 4.1 FE Update 3) includes libz.so.1.2.13.
To resolve this, I manually created a link named libz.so.1.2.3 using the makelink command. I couldn't find which process calls this library with that particular version so I created this workaround for myself.

Will work, of course, but it would help if we knew from where the MakeLink command giving the error is issued. Maybe you could run Snoopy after removing your self-created link?

rumbler wrote: Wed Mar 25, 2026 2:29 am MOUNT: "a" is not a valid keyword (file "*.info", line 1, column 1)

During system boot, the system attempts to mount files under DEVS:DosDrivers separately from their corresponding .info icons. Since .info files are not valid mountlists, this results in errors.

No, the standard Startup-Sequence contains this line: It means "Mount all the DOSDrivers files EXCEPT the ones ending in .info". If your Startup-Sequence doesn't have this exact line, it must have been modified. Or could this also have something to do with your special casing rules?

rumbler wrote: Wed Mar 25, 2026 2:29 am To work around this, I modified the S:Startup-Sequence file and added explicit checks for each device, for example:

IF EXISTS DEVS:DOSDRIVERS/PIPE
MOUNT QUIET DEVS:DOSDRIVERS/PIPE
ENDIF

This prevents the system from trying to mount .info files as mountlists.

Strange, because in effect it is exactly the same as the one-liner does which is supposed to be there (see above) - at least if you have included lines for each and all of the (non-icon) files in DOSDrivers.

rumbler wrote: Wed Mar 25, 2026 2:29 am Additional Observation

When enabling the Turkish language, I noticed that many translations are incomplete or missing.

Yes. For a number of languages, there is a lack of translators, or the ones that are still active, are alone on their language and have trouble keeping up (after all, it is a volunteer task like almost everything else related to OS4). Specifically for Turkish, I happen to know there is currently a translator, but he didn't have enough time before Update 3 to get all the missing translations done from many years without translation activity.

rumbler wrote: Wed Mar 25, 2026 2:29 am Also AmiSSL most recent version to the date I'm writing this message, install script has issues and can not proceed but as soon as I revert back to English, no problem...

I'm sure Oliver would like to know about this - oliver (at) amissl (dot) org.

rumbler wrote: Wed Mar 25, 2026 2:29 am You can likely reproduce this behavior easily in your own test environment, but if you need further information or would like me to test anything, I would be more than happy to assist.

Personally, I would be quite lost if I switched my system to Turkish. Let's hope someone who are familiar with the language can take it up from here.

Best regards,

Niels