Cloverleaf ads

I notice Cloverleaf are now placing ads. I’m not sure this is quite the wording they intended…

Strange :)

Hmmm, while we all want to see additional features in the OS I don’t think including mail order brides was on the list.
What next ‘happy endings’ massage?

Steve, you may well be proved wrong on that one!

And yes, we’d all like RISC OS to have a happy ending, many happy endings.

we’d all like RISC OS to have a happy ending, many happy endings.

Are we not talking more “out-comes” here?

It ain’t over ’til the fat lady …

Innuendo, out-she-comes!

Being more practical, maybe it might help with properly internationalising RISC OS…

I think it would take superhuman effort for us to outbreed the adherents of other OSes…

@ Theos

I notice Cloverleaf are now placing ads. I’m not sure this is quite the wording they intended…

Oh dear the end of the add is… quite miss-leading I would say lol But it seems more a security issue maybe?

@ Rick

Being more practical, maybe it might help with properly internationalising RISC OS…

Not a bad idea actually. BTW, thais reminds me of a little annoyance with internationalisation that still seems to on RISC OS (5 too). For example most of Run files still uses Error after an RMEnsure, not a biggie per se, but well we should add a simple tool to generate an error message that can be picked in the right language?

RMEnsure ModuleX 0.01 GenError 101

picks it from <Obey$Dir>.Resources.< Country >.Messages and make it part of the OS.

@ Clive

I think it would take superhuman effort for us to outbreed the adherents of other OSes…

+1, moreover with modern Open Source OS like Linux that effort is being perceived as “for free” by many…

IMHO:

There was a time for this back at the end of the 90s (but there were other issues back then). Now it’s probably all about slowly move forward with what it’s possible to be done.

Sure, adding new APIs to handle audio, video and accelerated 2D/3D graphics would be already a big achievement if done in a relatively short time, while ROOL works on adding the new Network Stack and WiFi/Bluetooth support.

And that is already a considerable effort in terms of design/coding/testing and make it stable because, as David J Ruck pointed well on another topic, RISC OS still requires far too many reboots (after hanging)…

BTW, thais reminds me of a little annoyance with internationalisation that still seems to on RISC OS (5 too). For example most of Run files still uses Error after an RMEnsure, not a biggie per se, but well we should add a simple tool to generate an error message that can be picked in the right language?

ResFind comes with the MultiError utility to do just this. There’s an example of its use in WinEd.

@ Steve

ResFind comes with the MultiError utility to do just this. There’s an example of its use in WinEd.

Nice! thanks for the info, I’ll have a look, but hummm if so then it sounds like we should advertise it more… Also ROOL DDE still uses just Error…

it sounds like we should advertise it more…

I only know about it because I’ve been working on WinEd… :)

ETA: Not quite true; I was aware of its existence, as ResFind is used in a lot of things and is pretty much the de-facto standard way to implement internationalisation in applications, AFAIK. I only investigated properly, and began to consider how broken my own apps were in this regard, as a result of working on WinEd.

ResFind comes with the MultiError utility to do just this.

I thought MultiError was an Adam Richards (Snowstone) thing and that the internationalisation element in ResFind was catered for by the ResError item in the source package of ResFind300.

Actually I think this revives the old thread about error messaging in the OS which ought to have certain features:

1. part of the base OS logging startup actions (see Harinezumi)
2. Multitasking – pop up an error notification (optional) and carry on (optionally stop the process). Martins “Reporter” may have something to contribute here as does the syslog utility regarding levels of logging detail.
3. Levels of logging detail – verbose¹ to terse
4. Log cycling to avoid filling storage
5. Archive to alternate media and/or zip compression as part of the log cycling
6. Internationalised messages

¹ Would extremely verbose be tagged R.M? :)

I thought MultiError was an Adam Richards (Snowstone) thing and that the internationalisation element in ResFind was catered for by the ResError item in the source package of ResFind300.

You’re right, I think. ResFind400 now, though, surely?

Incidentally, has the GAG site been restructured? I’m sure there was a partial English page for ResConf/ResFind at one point, but the links that I had seem to be dead.

Multitasking – pop up an error notification (optional) and carry on (optionally stop the process). Martins “Reporter” may have something to contribute here as does the syslog utility regarding levels of logging detail.

Not sure. The OS could really do with a standard way of logging errors, and that’s SysLog. Not least because one version of the OS already uses it.

Reporter is far more of a development tool, and I’d not expect production apps to log to it by default¹. Being able to divert a SysLog log to Reporter with a *command would be nice, though!

Levels of logging detail – verbose to terse

SysLog already does this, with provision for a granularity of 256 levels. WinEd adds a log_TEDIOUS beyond the standard constants defined by DeskLib for the purpose.

Log cycling to avoid filling storage

Doesn’t SysLog do this, or was that in the Other OS? Or am I just too used to Linux?

Archive to alternate media and/or zip compression as part of the log cycling

Again, SysLog (at least in some of its incarnations) could log to other hardware as I recall.

Internationalised messages

Obey files aside, there’s already support for this (and MultiError could plug that gap). In debug log messages, not so much, I think. That would be a significant amount of extra complexity, for (I suspect) very limited gain, given that log messages usually vanish if code isn’t compiled with DEBUG or similar defined somewhere.

¹ Let’s just ignore what WinEd does for now, eh?

Doesn’t SysLog do this, or was that in the Other OS?

Syslog tries to do this, but doesn’t always succeed. I remember back in the Iyonix days that some of the log files grew much larger than the supposed “maximum” size, requiring manual intervention.

You’re right, I think. ResFind400 now, though, surely?

I think the ResFind400 source is only the updated bits. Not looked closely so I could be wrong.

The OS could really do with a standard way of logging errors, and that’s SysLog. Not least because one version of the OS already uses it.

The other fork uses a version of syslog but even that version is only a slight update, syslog capabilities “out there in the real world” (Linux/Unix etc) have more capabilities.

The OS could really do with a standard way of logging errors, and that’s SysLog.

I was trying to avoid being prescriptive and just list features.

Again, SysLog (at least in some of its incarnations) could log to other hardware as I recall.

As I recall there’s an additional module kicking around (damned if a search of my archives can find it), certainly Theo’s WimpLog would be helpful for some features

SparkFS is using resfind since… for me since I can remember. StrongED also.
It is a international Territory app availabe … last German update was 5.18…

I think it would take superhuman effort for us to outbreed the adherents of other OSes…

Now, if there was a RISC OS bounty for that!

Now, if there was a RISC OS bounty for that!

It is my understanding that bonking for coin doesn’t tend to result in podlings…

Would extremely verbose be tagged R.M? :)

I see you’ve been spying on my debug output. ;-)

There’s a reason why software (such as Manga) has a global #define DEBUGCODE to enable the debugging functions, and then numerous #define DEBUG<THING> lines to selectively include what messages actually get baked into the app.

Of course, if we had a reliable debugger I might not need to embed ridiculous amounts of logging to trace what my code is doing, but since such seemingly simple things as DeskLib writing a value to file_lasterror can trip up DDT ¹², it’s often a lot easier to just include optional tracing while writing the code in the first place.

¹ It’ll work if one disables memory protection, but that has its own problems. And ironically DDT seems to reject writing to a code area (what I’m guessing DeskLib does?) yet never seems to fault null pointer accesses (unless &0 literally doesn’t exist).

² I wonder if it’s writing to a region marked as read only? I’ll need to check that. But if so, it should raise a warning, not an abort…

Rick, I’ll throw in the bonking for free, I’m thinking more of a contribution to the podlings considerable running costs once hatched.

It is my understanding that bonking for coin doesn’t tend to result in podlings…

“happy endings” is more of a ‘manual handling’ issue.

I see you’ve been spying on my debug output.

Just a talent for reaching useful conclusions from minimal data. :)