Need Feedback for (wip) UHC-GUI

[EctoOne]

Originally Posted by DisasterIncarna

nice, yeah i was thinking of any future updates for the GUI, wasnt sure if the slow increases in size would eventually be an issue or not.

I'm currently rewriting the guide, which will end up being bigger without the feature I want to add to it. I will see how much space I'll have left before adding anything new. And if space is running low I'll think about something. But with the last version I was able to also put JanoEditor on the disk, so there should be some space left. Especially since I've not added it to the install script.

I've also spent some time messing around with the guide, and it is possible to run commands/scripts from a guide. So, I can basically can create a list of recommendations which can be downloaded with a simple click. If space gets too low, I might just remove all optional files from the install script and put them in the guide.

Edit: Dang, rewriting the guide was already too much. The disk got full during the installation of the GUI. So, I have no choice but to delete the included packages after they got installed. Still need to check the free space after a full installation.

[patrik]

Originally Posted by EctoOne

Yeah I've had that socket error like forever. I believe it's emulator based.

Which emulator are you using and do you have an example of what to do to reproduce it - preferably using as little software as possible to reproduce it?

Have seen many issues with the bsdsocket.library implementation in various UAE variants and already have a few workarounds in the code for UAE issues. It would be great if you could bug report it to the emulator authors.

Originally Posted by DisasterIncarna

i got it when the gui called aminetrecent, the error would be latched onto the end of the output and the gui mistakenly thought it was a valid entry so i just skip past anything that starts with that error message, plus it only happens on steam deck emulation for me, never happened on my main pc.

Does it also happen when you call aminetrecent in the shell? Would be great if it could be bug reported to the emulator author.

[EctoOne]

Originally Posted by patrik

Does it also happen when you call aminetrecent in the shell? Would be great if it could be bug reported to the emulator author.

I'm getting the following when running aminetrecent in the shell on uae4arm on Android (the playstore version)

setsockopt(): Numerical argument out of domain (33)

Sadly I don't think that there is much development on the emulator. There is a newer version on github, but that version never really worked for me.

[Swe_Kryten2x4b]

Originally Posted by DisasterIncarna

If someone can make sure there is no delete error message on readmes now then i will again update the first post archive with this test version.

cheers.

Works for me. I've tried with Aminet, Aros, and MorphOS with no error messages. And several readme files in the recent-list for each site.

So seems to work as it should.

[patrik]

Originally Posted by EctoOne

I'm getting the following when running aminetrecent in the shell on uae4arm on Android (the playstore version)

setsockopt(): Numerical argument out of domain (33)

Sadly I don't think that there is much development on the emulator. There is a newer version on github, but that version never really worked for me.

I am using setsockopt() to set a receive timeout - setsockopt(…, SOL_SOCKET, SO_RCVTIMEO, …). A not that uncommon operation. SOL_SOCKET and SO_RCVTIMEO are actually numbers defined in the bsdsocket.library SDK and my guess is that the emulator author has missed that he needs to translate them to the domain of the setsockopt() in the host OS (Android/Linux), which is what will be called in the end.

You should also get it from any of the other search or recent scripts as they call the same UHCSearch binary.

[patrik]

Originally Posted by Swe_Kryten2x4b

Works for me. I've tried with Aminet, Aros, and MorphOS with no error messages. And several readme files in the recent-list for each site.

So seems to work as it should.

Yes, have not seen this issue myself. Tested on all Amiga TCP/IP stacks, winuae, fs-uae and e-uae. What do you run?

[DisasterIncarna]

Originally Posted by patrik

Which emulator are you using and do you have an example of what to do to reproduce it - preferably using as little software as possible to reproduce it?

Have seen many issues with the bsdsocket.library implementation in various UAE variants and already have a few workarounds in the code for UAE issues. It would be great if you could bug report it to the emulator authors.



Does it also happen when you call aminetrecent in the shell? Would be great if it could be bug reported to the emulator author.

its latest 3x FS-UAE for Linux which im using on the Steam Deck, as i did it via the GUI and nothing seemed to break, i just kind of hid the error, ill try and reproduce it on the steam deck once a few accessories arrive later today, a dock and a better bluetooth keyboard so its less of a nightmare using desktops and i can try a bigger range of functiosn without using the GUI and see what errors pop up.

For me this is only happening on the x86 linux build, ive only seen this on the steam deck, so maybe its the emulator, maybe its the version of linux valve uses, would need to try the linux build on different linux distros to know more, i get no such error when i compile fs-uae for arm for use on the pi400, nor have i had the error on the windows version.

[Swe_Kryten2x4b]

Originally Posted by patrik

Yes, have not seen this issue myself. Tested on all Amiga TCP/IP stacks, winuae, fs-uae and e-uae. What do you run?

A1200 with kickstart/wb 3.2.1 and Miami Deluxe. And, according to uhcupdate, the latest version of the tools. The GUI is started from Amidock in cli-mode.

But since the latest test-version of the gui fixed it the issue must have been that >nil:

[patrik]

Originally Posted by DisasterIncarna

its latest 3x FS-UAE for Linux which im using on the Steam Deck, as i did it via the GUI and nothing seemed to break, i just kind of hid the error, ill try and reproduce it on the steam deck once a few accessories arrive later today, a dock and a better bluetooth keyboard so its less of a nightmare using desktops and i can try a bigger range of functiosn without using the GUI and see what errors pop up.

For me this is only happening on the x86 linux build, ive only seen this on the steam deck, so maybe its the emulator, maybe its the version of linux valve uses, would need to try the linux build on different linux distros to know more, i get no such error when i compile fs-uae for arm for use on the pi400, nor have i had the error on the windows version.

Sounds like the Linux version is the culprit except that it works when you compile it for the pi400 . It would make somewhat sense as the numbers used for bsdsocket.library is for bsd derived tcp/ip stacks, which windows and os x is, but Linux isn’t so a translation is most likely needed in that case in fs-uae’s bsdsocket.library implementation. Still, why does it work on rpi400 .

Have you tried fs-uae on a normal linux desktop?

[patrik]

Originally Posted by Swe_Kryten2x4b

A1200 with kickstart/wb 3.2.1 and Miami Deluxe. And, according to uhcupdate, the latest version of the tools. The GUI is started from Amidock in cli-mode.

But since the latest test-version of the gui fixed it the issue must have been that >nil:

Have not seen any functional issues with Miami.

The >NIL: I believe was to silence the output of the delete command when files were missing. If you redirected the search to NIL, you would not see any results.

[Swe_Kryten2x4b]

Originally Posted by patrik

The >NIL: I believe was to silence the output of the delete command when files were missing. If you redirected the search to NIL, you would not see any results.

Well, yes. And that was the only "problem" I had. The unneccessary output window that is.

[patrik]

Originally Posted by Swe_Kryten2x4b

Well, yes. And that was the only "problem" I had. The unneccessary output window that is.

Thanks for confirming!

[DisasterIncarna]

i havent tried on a normal linux distro yet, i only have 1 "normal pc" left that runs winblows and i didnt want to dual boot or the likes incase as usual, something catastrophic and unexpected happened so i left linux to alt devices like the pi400 and an old notebook (which hasnt really got the power to run an emulated amiga so havent tried).

Might try the old acer aspire one notebook even if it is slow, it has an old and out of date linux that refuses to update so might have to overwrite the whole thing with a modern/current linux distro, even slow at least i can confirm or rule out and narrow down any issue with bsdsocket.

[DisasterIncarna]

note: the version of UHC-GUI that just appeared on Aminet is in fact 1 version behind the test version that sent delete messages to NIL: so if you have the test version, keep using it.

[EctoOne]

Originally Posted by DisasterIncarna

note: the version of UHC-GUI that just appeared on Aminet is in fact 1 version behind the test version that sent delete messages to NIL: so if you have the test version, keep using it.

Is the version available on Aminet the one which hides the socket errors? I've almost finished a new version of the disk and noticed that the errors were gone and I thought it was caused by the unusual start of UHC on the disk until I remembered reading something about hiding the errors.

[DisasterIncarna]

Originally Posted by EctoOne

Is the version available on Aminet the one which hides the socket errors? I've almost finished a new version of the disk and noticed that the errors were gone and I thought it was caused by the unusual start of UHC on the disk until I remembered reading something about hiding the errors.

no the socket error version still needs to be uploaded, i was just waiting for more feedback before sending that version, as nobody seems to have an issue with the test version....so far, ill send that version now, until then keep using the test version.

[EctoOne]

Originally Posted by DisasterIncarna

no the socket error version still needs to be uploaded, i was just waiting for more feedback before sending that version, as nobody seems to have an issue with the test version....so far, ill send that version now, until then keep using the test version.

Well, that's weird then. Because I don't get the errors on my disk installation.

Here is the newest version, again rename the file from UHC-Disk.zip to UHC-Disk.adf

I've changed some things, mainly I moved the GUI directly into the UHC directory. Theoretically it should be possible to drag and drop the directory to a new location now. It still needs the disk though, because I've not yet worked on a real HDD installation. That would be the next step and actually should only require to copy LhA, System and WBRun (with the parm.library) to a HDD. But I have not tested that and obviously want to add it to the install script.

I'm also curious if it is clear enough how to even start the installation now.

[DisasterIncarna]

new version up on first page, UHC devs added support to Search/Get/Extract from Turrans FTP (Grandis.nu), they warn at present search results can be LOOOOOONG and wont all fit in the window, future version of the gui will "try" and trim that down to fit as best as possible, if in doubt tho click any entry and just like previously a trimmed/more full set of details is shown in the info bar at the bottom which uses a non fixed width font so you "should" see all details that way for now.

Also to note, the RAW filename given is based off the FTP's way of doing things, ie: filenames that amigaos/workbench doesnt like might clusterfudge when trying to download or extract, if an EXTRACT operation fails, try GET, and even then check the resulting filename if it worked, it may need a quick rename, this is still a work in progress at a guess and the GUI wont need a fix once the UHC guys find a way around the filename issue. Files with more "sensible" names are ofc no issue at all.

As an example, try TURRANSEARCH SIM CITY, then TURRANEXTRACT 0, or just type SIM CITY into the GUI for Turran FTP, it will list a lot and if you look at the first entry, the filename has + symbols, square brackets [], loads of spaces in it, and the whole thing kinda just facemelts at the present with a filename like that, this will no doubt get a fix later, just do periodic checks for updates to UHC-TOOLS via the GUI or type "UHCUPDATE" in a shell window.

[Swe_Kryten2x4b]

Very nice! I did a quick test and for a first version with that functionality, it works a treat.

I only tried to search for populous in the GUI and while it didn't all fit in the window it was easy enough to see and use.

[DisasterIncarna]

Originally Posted by Swe_Kryten2x4b

Very nice! I did a quick test and for a first version with that functionality, it works a treat.

I only tried to search for populous in the GUI and while it didn't all fit in the window it was easy enough to see and use.

if it doesnt fit, which is likely, try clicking the entry and look at the info bar at the bottom, that uses a proportional font and should squash more text into it.