Patience is a nice virtue

Hi folks,

While I rewatched the Millennium trilogy in between FreeBSD source rebuilds and remembering how I appreciated Rapace portrayed the emotionally restrained and troubled nature of the character, I finally got some confirmation for contributions back from 2018's autumn. For instance, memcached finally merged in one shot two of them related to FreeBSD I mentioned at the time ... the super page support making this OS equal to Linux in this area ... and also using basic capsicum capabilities for the "sandbox mode" (aka drop privileges here).
Also finally redis, after a well deserved long pause, started to accept merging again including the BSD build fix. Since both projects are used in my day job, that comes to handy :-) Professionally speaking, I updated the HAPoxy DeviceAtlas module this week in order to fit better the future 2.0 version requirements.
Once my system is fully up to date, I will see which projects I can help out within the weekend, we shall see ...
Noteworthy if you re interested in LMMS status, there is this topic here you can overwatch, giving good overview what is going to happen...  Also for broader plugins support there is this one. In the meantime I wish you all a great weekend !

You are early again !!

Hi folks !!

The spring here goes smoothly, a bit rainy as usual in this country but temperature wise it is getting better :-) This week had witnessed the early release of OpenBSD 6.5 but not too far from the 1st of May this time. As you can see most of the packages are really up to date (despite the high level of patching Chromium is pretty current, for web development PHP 5.6.x had been dropped now only providing PHP 7.x, logically following up the end of life time for this version branch, Golang which has the last release, ...) but system-wise it has also its share of updates :
- unveil underwent some changes, it has been the main focus of improvement, as pledge was in the past, in order to fit program consumers for accessing configuration files and so on.
- kubsan to have decent undefined behaviour code coverage, the checks subset available (especially overflow/underflow and out of bounds) is good enough to find diverse types of subtle bugs. Very useful and glad to see the efforts put behind.
- clang on mips64 ... amazed how they achieve to push down the obstacles, architecture-wise, release after release :-)
- As a developer, interesting to see malloc now uses sysctl.
There are many more of course, depends where your own interests lie.

Todd Mortimer had updated his paper used for the last AsiaBSDCon ... Still the videos not available but the content of this very interesting talk can help to understand the importance of his work.

Toured again recently into some useful open source projects which I used recently :
- Google bloaty ... I needed to check binaries size efficiency and helped me to understand better how some were so important and how to reduce them eventually.
- opencv which I already mention I believe ... I used for picture treatments and happens to be (genuinely) Android friendly.
- Otherwise PHP Parallels ... To be able to do concurrent programming with PHP is quite something ... and it comes from krakjoe who is one of the most important php contributors, maybe one of these days it will be integrated upstream just like his phpdebug project, who knows...

Speaking of PHP, a lot of efforts are put towards the JIT for opcache feature for the next 8 version. dstogov, the main developer is very efficient in Linux development so I just need to check time to time if it works all right on BSD and fix it when necessary ; but usually it is a matter of few lines so it is very good considering the level of complexity ...
These last weeks I have been into botan a little bit, most notably improving slightly freebsd support for detecting cpu features and also using more reliable way for the thread pools creation (I had Android in mind at this moment). Otherwise a bit of usual radare fixes here and there and some other changes are waiting approvals.

Out of memory or not ... this is the question !!

Hi folks,

Not questioning here your memorisation capacities here :-) even though in our modern days we have so much informations to integrate ... nevertheless this week had seen a long waited release of the famous memory allocator jemalloc, 5.2.0 for instance. As you read the changelog, you can feel the months of development. I personally appreciate the possibility to disable the dl library usage, the rework of malloc/free fast path but above all trasz and jasone (even just the continuous integration work alone is already a great achievement) had done a great job on improving the FreeBSD support so thanks to both ! My only little modest contribution for this release was to enable unit tests on FreeBSD, Ed Maste was needing this at some point but nevertheless it is always nice to increase balance b/w operating systems whenever we can. Another nice memory allocator I mentioned a bit few months ago, rpmalloc, still evolves over time, after having improved slightly the BSD support I have been pushing little change this week .

Before my memory fails me, radare had finally gotten its 3.4.x release (hufflepuff) with as usual a lot of internal rewrites, bug fixes and new features. Couple of pull requests I have pushed in the previous week, we shall see if some will be merged ...

Also, OpenBSD will have soon its 6.5 release and on the mailing list we can see time to time this kind of messages ... It has happened in the past and most of the time the poster has a good point but this is more the way the message is conveyed ; but in this case I have to agree with other OpenBSD people, updating redis from 4 to 5 so close to the release is too risky, there are users and also enterprises using this os, there is a good chance some packages would break as the redis project is pretty active that I can guarantee ... Do not forget it is an open source project with people doing this in their spare time ;-) do not let the anger gets the best of you.