Running the app as regular user outputs wrong read speeds

[raspin0]

• Linux-distro (kernel version): Deepin 20.7.1 x86_64 5.18.17-amd64-desktop-hwe
• Desktop Environment (KDE/GNOME etc.): Deepin
• Qt Version: -
• KDiskMark Version: 3.1.2
• FIO Version: fio-3.32

Description:

Flatpak & Appimage (NO ROOT ACCESS)

Appimage (RUNNING AS ROOT) - Correct results (Compared against other benchmarks of this pendrive)
Command: sudo env DISPLAY=$DISPLAY XAUTHORITY=$XAUTHORITY ./KDiskMark-3.1.2-fio-3.32-x86_64.AppImage

[JonMagon]

That's why a special warning is displayed at the top right.

[raspin0]

The warning does not inform you that the readings will be incorrect, nor it doesn't specify that root access is required to perform a proper benchmark. If the application doesn't work on Flatpak and Appimage and there is no desire to fix issues why builds of this application are available on those platforms?

[JonMagon]

It does. Whether it is right or wrong depends on the device and the mounting arguments.

[raspin0]

Where does it say that the application may produce incorrect results? I don't think the app should launch at all if there are such known problems.

[JonMagon]

Right at the beginning.

[raspin0]

That's not true, limited functionality doesn't mean or even imply that the application may produce incorrect results.

[raspin0]

I guess there's no point in arguing like that, so I will just leave my final opinion.

Running KdiskMark in usermode as a Flatpak and Appimage is flawed and it's known to produce incorrect values, hence unless there's a will to fix the issues, it should be forbidden to run without root and removed from platforms where it doesn't work correctly. Any kind of warning, even a bigger one is not enough. One of the core principles of such tools is reliability and without strong focus on quality of produced results I personally can’t recommend this application to anyone.

[JonMagon]

I say it again: everything works as expected.
If the caching distorts the result, the user needs to run the portable app as a super user to be able to bypass it.
Running the GUI as root user is not the best practice, but the only possible one in this case.
Moreover, it is usually the internal devices that are tested. Or right configured external ones.

Whether or not to recommend it is your decision. Flatpak and AppImage packages are successfully used by many people.
I'm not going to limit the ways of delivery, or to force the user to run an app from the Internet as sudo.
The warning in the upper right corner is sufficient.

[raspin0]

How am I supposed to know when the caching issue occurs or even exists? The warning is not self-explanatory at all. If I wasn't techy enough, didn't have a comparison benchmark and trusted this program I might have trusted the result.

Whether or not to recommend it is your decision. Flatpak and AppImage packages are successfully used by many people.
I'm not going to limit the ways of delivery, or to force the user to run an app from the Internet as sudo.
The warning in the upper right corner is sufficient.

It's your decision, but not realising this as an issue and knowingly showing users incorrect values isn't nice and heavily undermines applications general reliability.

It's like you don't want users to run apps from the internet as sudo, but realistically if a normal user just wants to have a functional application and proper results it must run it as administrator.

[JonMagon]

Ok, I will expand on the warning text.

[raspin0]

Thank you for understanding. That is a very good idea. I'd think about ways of actually fixing this issue or if it's unfixable just making sure there's no way a user can get an incorrect values. Even if the user skips all the warnings there should be a system in place that prevents incorrect values from ever appearing.

Ultimately though, imo. it's unacceptable to have an app that half works dependingly on where and how you turn it on.

Good Luck :)