Notices by Romain Tartière 😈 (smortex@mamot.fr), page 5

@kazord "au bon vouloir de l'implémentation" car au détour de https://pubs.opengroup.org/onlinepubs/009695399/functions/fprintf.html on peut lire: « The low-order digit shall be rounded in an implementation-defined manner. »

@kazord
Moralité:
double value = ...;
double multiplier = pow(10, precision);
double res = round(value * multiplier) / multiplier;
printf("%.*f", precision, res);

@kazord Bien, j'en suis venu à la même conclusion: printf(3) arrondie n'importe comment au bon vouloir de l'implémentation, et round(3) fait le boulot correctement.

En effet, 0.05 c'est pas un a×2^b (donc coup de chance, il s’arrondit à 0.1), 0.25 est est un (1×2^-2) est printf("%.1f", 0.25) retourne "0.2" qui est mathématiquement faux mais est cohérent avec ce qui a été vu ci-dessus.

@Nervengift @tom Awesome! Do you have a .svg / .pdf for sharing this?

@mhka I always wanted live-systems to rely on this kind of mechanism…

This way, you:
Boot on the live-cd / live-usb-stick;
Use the product, be happy, decide to keep it;
Run the installation process (which does what you show);
Remove the device you booted from to free the CD-ROM drive / USB port;
All work you did in the live system is *not* lost;
All customization you did in the live system is *not* lost;
No need to reboot.

@hirojin because we trend to remember them and are unable to invalidate our cache properly?

@kazord et puis, justement, 0.5 c'est 1 × 2^-1 🤔 😥

@kazord j'y ai pensé, mais roundf(3) arrondie comme espéré… Du coup je n'arrive pas a capter ce qui fait que dans un cas ça arrondie au dessus et dans un autre au dessous 😕

Of course:

printf("%.1f", 0.05);

outputs 0.1, not 0.0.

Fun with C:

printf("%.0f", 0.5);

outputs 0, not 1.

#rounding, #c

@jpmens @dch and I use #riemann on FreeBSD, which I think deserves more attention… I am thinking about preparing a talk on it in order to help it gain more visibility, but I don't really know the audience I should target, so maybe the ideas you have might help me in some way?

cc @faxmodem who also use Riemann (but not on *BSD AFAIK)

@raichoo Found it! Thanks!

https://hub.darcs.net/raichoo/hikari/patch/0e6d64f247a02e5a425e5f81832087c5c517b0a8

Isn't `zpool status` supposed to show "(resilvering)" in front of the disk that I have just added to a pool?

I do have information about a resilver in progress in the header, but expected to see "(resilvering)" at the end of the line of the device (gpt/disk0) I added to the pool (using `zpool attach zroot gpt/disk1 gpt/disk0`).

I feel confused 🤔…

#FreeBSD 12.0 #zfs

@saper Hehe, probably, but I am not a Java developer and was having difficulties finding the appropriate tools 😉

Some people helped and I finally found the problem in some C code that was not properly handling the communication with the clojure server, causing it to wait forever for the client to ACK before releasing memory…

Committed to the FreeBSD ports tree!

@hirojin Sure it affects us — in my case in a good way — but I was already a happy contributor before. I am just a bit happier now 🙃

Asynchronous communication helps a lot here to continue to contribute efficiently, and I guess that my move does not have any impact on the quantity of contribution I do.

Also, better living conditions (climate, location, people) are a good way to be globally 'happier".

So, maybe it would have been wise to leave earlier 😋