koitsu wrote:
This thread is a guaranteed slow-motion train wreck. Nobody can definitively tell you what's best, because that's so incredibly subjective that it's almost pointless to ask.
Hence why this was asked on a forum, not a Q&A site. I read it as "Of the distributions you have tried, what about it worked for you and what did not?" But my Q&A site habits are still leaking out my fingertips, so let me get this one out of the way first:
"My partition for desktop GNU/Linux is too small. How can I choose a distribution to make it bigger?"
You may have hit an
XY problem. This means you ask for help with details of a particular solution when a completely different class of solution that you had not anticipated may suit your underlying problem better.
If you're running out of space in general, my first recommendation is to run
du on your home directory to figure out where most of your stuff is stored. If system directories outweigh your home directory, that means your partition might be too small. But otherwise, you can move large files to another partition or to external mass storage. Or you can clear various caches, such as that of your web browser or running
make clean on your programming projects.
But with that said, if you're stuck on a relatively small partition, such as if you're trying to make a desktop GNU/Linux distribution share a laptop's small SSD with the Windows 10 that came with the laptop, I recommend Xubuntu. Its Xfce desktop avoids the excess bloat of things like Unity (default in 11.10 through 16.04) and GNOME 3 (beginning in 18.04), and it may fit into a smaller partition. But I have no first-hand experience using recent desktop Debian or Ubuntu in a partition smaller than about 50 GB, though I've used Ubuntu 8.04 (with GNOME 2) on a netbook's 4 GB SSD before.
As for just using Windows 10 on the desktop and GNU/Linux on the server, one of the advantages of GNU/Linux on a software development workstation is that you don't need to connect to the Internet or carry a Raspberry Pi around with you to test server code that you write that exceeds the capability of WSL. Nor do you waste as much time on the staging server working around differences between the Windows and UNIX versions of your favorite server programming language's standard library, which would be the case if you test on Perl, Python, PHP, or Node for Windows and deploy on Perl, Python, PHP, or Node for GNU/Linux. Nor does testing graphical applications require downloading a ten-year-old unmaintained copy of Xming, which you would have to do on Windows because WSL lacks an X server.
In addition, Windows 10 tends to be a bandwidth hog around update time. It gives the end user little or no notice that it is about to download a semiannual upgrade and thereby cause the ISP to bill the subscriber for monthly data transfer quota overages, not to mention the loss of unsaved state that an unattended restart represents. (If there were more notice, the user could drive the computer into town and perform the upgrade on the county library's unmetered Internet.) Unlike Windows 10, Ubuntu gives about 3 months of notice for the semiannual upgrade track, and the 2-year upgrade track of Debian and Ubuntu does not carry a surcharge.
As for just using Windows 7 on the desktop, the security updates for Windows 7 are due to end in two years, and a newly purchased laptop will have neither the drivers nor the downgrade rights to run old Windows.
koitsu wrote:
If [avoiding Poetterix] doesn't matter to you, then great, but if you ever have to deal with it on even a general level (like
dealing with the journal getting corrupted -- I sure do love Lennart Poeterring's reply!)
I love it too because I see his point. I don't see the problem with using EOF as a marker to contain the corruption at write time. This way, later versions of the reading program can include improved error correction without having to correct for both the original error and the old version of the reading program's attempt to fix it.
koitsu wrote:
I still can't take Gentoo seriously (the arguments for it are always "but Portage is awesome", to which I reply: I've used FreeBSD since 1997, where do you think they got the idea from? Ports/Portage ain't enough to justify use of an OS, sorry).
I guess the rationale is the ports tree of FreeBSD combined with the hardware driver support of Linux.