Profile cover photo
Profile photo
Sriram Ramkrishna (sri)
sysadmin, open source, GNOME, GNOME Community guy, photographer, a man who can hold his breath for 15 minutes
sysadmin, open source, GNOME, GNOME Community guy, photographer, a man who can hold his breath for 15 minutes


So I removed all the spam, and then y'all don't post anything for 14 weeks! What the hell! :D

I'm really excited for this year's GUADEC which will have a number of downstreams. I reckon we are going to see folks from elementary, purism, Endless and system76. In addition of course to openSUSE, Fedora, and Ubuntu.. This is a good thing as more representation from downstream will help sensitive GNOME to how downstreams use their software.

While GNOME rightly thinks of itself as a product, downstreams also are deriving from the GNOME platform and it is worth encouraging both participation and listening because both will encourage more resources in the platform that GNOME desperately needs.
Add a comment...

I have removed all the spam posts and used my ban hammer and did some damage. Hopefully that will help some. Thanks for bringing this to my attention.

Post has attachment

Post has shared content
If I don't see any people from the Barcelona at GUADEC, I would be disappointed! (GUADEC is in Spain this year)
City of Barcelona wants to dump Windows and Microsoft's products and move to Ubuntu Linux, LibreOffice, and Open Source software!
Add a comment...

Post has attachment
I found this fun rant while going through GNOME's twitter messages. (I'm trying to make an effort to spend a little time on it every couple of days since I'm sort of doing the same tasks for System76 and Pop!_OS)

The author laments the end of menus and various other UI bits. Brings up the great +Andy Hertzfeld especially in regards to the nautilus changes. It seems he has retreated to Mate to hide from the changes being done across all the linux desktops.

I don't particularly have a profound statement to make on this. GNOME designers (and Elementary as well) don't like menus. Unity takes this even further with the lens feature where you can just search the menu item from the app. Ultimately, menus are problematic because they require precision to find and choose and there are better ways to deal with them. They especially make it difficult in the hybrid world of 2-in-1 laptops.
Add a comment...

I noticed that Linux Plumbers is cohosted with Kernel Summit. While understandable that you want to co-host, but it reinforces the perception that LPC is nothing more than an extended kernel conference with a wider audience. If you're going to care about the entire eco-system then it needs to be more than just hardware and kernel.

Taking a look at last year's schedule, it seems that we've decided that userspace is not at all important in the scheme of anything. I guess desktops and applications are no longer something that is deemed part of the Linux eco-system.

Since the rest of us are left to fend for ourselves, I am sure we can fill up the gap with our own conference and work to increase that 4% desktop market share. But I'll happy invite our kernel developer friends to the party though.

Add a comment...

My way of dealing with non-GNOME fanboys now is to point them to monetary contribution pages of their favorite DE and ask them to donate. I figure that's way more positive than trying to engage with them in any other way. After all, giving resources to desktop projects is a good thing of whatever type.
Add a comment...

Post has shared content
I wonder if we need to start looking into this with GNOME stuff.
So then there's the issue of detecting attempts at exploitation. Let's pick Meltdown for a start.

There are multiple ways to implement an exploit, but most cache side channel attacks of this type will rely upon tight timing loops doing an rdtsc/rdtscp loop sequence with an embedded memory access in the middle. That's not quite unique enough for runtime, but you could modify virus scanners, script runtimes, or binfmt handlers to look for these in combination with cache flushes.

...and also...

In the common flush+reload type approach, you can use performance counters to measure the cache miss rate at run time, while in a flush+flush approach, you can't but you can look in the binary images for lots of clflush instructions and suspect sequences of clflush, prefetch<number>, and the PMU indications of the cache activity.

This is something I pondered aloud before the holiday with some colleagues. I am now interested in just sharing that we thought about it (and some other things I won't share just yet) and that I am open to feedback on whether there is any value in writing this up into some kind of proposal.

I would bet money some of the app stores are already doing some binary analysis.
Add a comment...
Wait while more posts are being loaded