Centurion_Dan | So I upgraded my laptop to beowulf, and I like it a lot. Everything works. All my work to support grub themes finally has paid off as beowulf finally has grub theme support making it look real sweet at the bootloader, and xfce4 doesn't barf whenever I change plugin/unplug monitors. The only thing it's missing is support for monitor profiles which is in the next release of xfce4. | 09:39 |
---|---|---|
Centurion_Dan | overall I'm really impressed with the monitor. | 09:40 |
Centurion_Dan | s/monitor/beowulf upgrade/ | 09:40 |
xrogaan | yay | 09:58 |
blstrb | do you put ntpdate -b server.name in /etc/rc.local? | 13:25 |
r3boot | depends. If ntpd is already running by then, you cannot allocate src port 123, and ntpdate will fail | 13:26 |
blstrb | installing ntpdate doesn't put a startup in /etc/rc2.d (my default runlevel) | 13:28 |
r3boot | Correct, since ntpdate is meant to run just once. Normally, you'd run ntpd which will keep time time in sync | 13:28 |
blstrb | i don't seem to have ntpd | 13:29 |
blstrb | ntpdate - client for setting system time from NTP servers (deprecated) | 13:29 |
blstrb | what should i install instead? | 13:29 |
r3boot | ntp; once that's installed, configure /etc/ntp.conf with 4 NTP servers (stick to <yourcountrycode>.pool.ntp.org), and keep that running in the background. | 13:31 |
r3boot | that will keep your time in sync; with ntpdate, you will need to run it from cron to keep your machine synced over time | 13:32 |
r3boot | but, dont deinstall ntpdate, since you need that to solve a minor inconvenience with ntp, namely, that it will doe stepping if your time is off, while ntpdate -b will force the time to the current NTP time | 13:32 |
r3boot | Eg, normal operation: run ntpd; To fix a large time offset, stop ntpd, ntpdate -b, start ntpd | 13:33 |
r3boot | blstrb: see https://maze.io/paste/p/YypGbVUQrJ0FRNpc as an example for ntp.conf | 13:36 |
blstrb | thanks r3boot | 13:39 |
blstrb | i still have habits from the 80s and 90s | 13:43 |
r3boot | ghe :) | 13:43 |
* r3boot was an ntp operator for approx 15yrs, but gave up after reflection became a hot new thing (again) | 13:43 | |
blstrb | what amount of clock drift will break https for browser? | 13:47 |
blstrb | this is really offtopic but you might know offhand | 13:47 |
r3boot | afaik thats somewhere between 1hr and 24hrs, depending on the browser. But systems that clockdrift that much usually have hardware issues which need to be solved | 13:48 |
r3boot | not that it really matters, worse case you can ntpdate this | 13:48 |
r3boot | I used to have this AlphaServer which had a rotten RTC, and it would run 1.2s in the time where reality runs in 1s. ntpd was not able to keep up with it, but * * * * * /usr/sbin/ntpdate -b in cron fixed that more or less | 13:49 |
r3boot | nasty, and eventually I replaced this box for something better | 13:49 |
blstrb | wow | 13:50 |
r3boot | On x86(_64) I havent seen a lot of issues in the last 10yrs or so (got experience with dell/hp/lenovo servers) | 13:50 |
blstrb | my company moved the business logic from 300 pcs to one alpha server emulating x86 and it ran like 20x faster | 13:50 |
r3boot | hehehe, sounds expected :) | 13:50 |
r3boot | Alpha can be fast, if your compilers can generate proper code for it | 13:51 |
r3boot | .. and that already exludes gcc for most part :/ | 13:51 |
blstrb | ---------------------------------+ | 13:53 |
bob38 | list | 14:11 |
avbox | FYI, there is a new AVMultimedia release, it is kernel 5.3.6 based. Supports now bluetooth and was widely tested on intel compute stick (BOXSTK1AW32SC). Setup in 20 to 30 seconds, on fast machines even in 10 seconds. If you have questions, just ask. | 18:41 |
EHeM | The situation with the Apache2 packages is kind of serious given how that is a *security* issue... | 19:24 |
buZz | whats the context? | 19:25 |
EHeM | All of the amd64 variants of deb9u9 are missing. | 19:27 |
mason | Random note for anyone mucking with custom installs - my early ZOL 0.8 builds were using elilo, but it turns out that the kernel Debian ships is happy being an EFI stub, so now I'm booting a native-encrypted, TRIM-capable ZFS mirror as root, without bothering with a bootloader at alll. | 19:29 |
EHeM | https://pkginfo.devuan.org/ strictly reports information from amd64, so if you've got a amd64 machine which also loads the i386 repositories it will report wanting to upgrade apache2-data; guessing that is due to apache2-data being all and thus the deb9u9 package from the i386 repository looks acceptable. | 19:29 |
buZz | EHeM: are you talking about buster? | 19:32 |
EHeM | buZz: Ascii/stable. | 19:32 |
buZz | oh ok | 19:33 |
buZz | you could pull the newer ones from backports it seems? | 19:33 |
buZz | oh thats just some libs, nmind | 19:33 |
EHeM | I wonder whether the core of people remaining with current Debian would be up for the task of trying to push various packages to start making use of `libtool` and making libraries optional, it is extremely difficult to setup a full desktop with graphics without installing *both* libmysql and libpq since many chains of library dependancies eventually pull both in. | 19:41 |
EHeM | On servers often more than 50% of installed packages are libraries which `libtool` could change to recommended instead of depends. | 19:44 |
buZz | EHeM: there is 'equivs' to prevent installing deps you dont want | 19:46 |
EHeM | buZz: Doesn't work for most libraries as the program needs to be using dlopen()/dlclose()/dlsym() instead of doing normal linking. | 19:47 |
furrywolf | personally, I'd like to see a change to how programs are written in general to make most libraries optional, as long as you don't need the feature the library provides. that is, at runtime missing libraries are ignored rather than causing the program not to load, then the program later checks to see if a library was loaded before trying to use functions it provides. | 19:49 |
buZz | EHeM: hmhm | 19:50 |
djph | furrywolf: usually you do stuff like that at compile time | 19:55 |
furrywolf | this is something that's already doable with loading libraries at runtime, but it's annoying enough that no one does it. | 19:55 |
furrywolf | djph: yes, I know what you usually do. that's why I said it would be a _change_. :P | 19:56 |
djph | furrywolf: but we already have a system that works | 19:56 |
djph | furrywolf: don't you go try systemd-ifying compilation now too | 19:56 |
furrywolf | try loading the library at runtime instead of directly linking against it, then replace all your #ifdefs with-foo with if(isloaded(foo)) type checks... | 19:57 |
furrywolf | we have a system that makes you install libpulse to run gimp. | 19:57 |
furrywolf | the system clearly has room for improvement. :) | 19:57 |
djph | well, anything that installs pulse has room for improvement | 19:58 |
furrywolf | very true. but in this case, it's a program with absolutely no audio support or awareness whatsoever, that ends up requiring it through a library that uses a library type thing. | 19:59 |
furrywolf | I suspect that less than 50% of library dependencies ever have a single function used from them on any given install. | 20:00 |
EHeM | furrywolf: More like in 75% of installs the given library is never called, but it is loaded due to the runtime linker. | 21:36 |
EHeM | furrywolf: The mechanism already exists, dlopen()/dlsym()/dlclose(), problem is it is something of a pain to use (libtool tries to take care of the worst issues) and thus many developers never bother. | 21:37 |
furrywolf | yep. | 21:37 |
EHeM | Notably some systems the standard is include an underscore prefix on symbols, which means on those systems you need dlsym(handle, "_XGetXCBConnection") instead of dlsym(handle, "XGetXCBConnection"). | 21:39 |
EHeM | Some systems/runtime will call the function _init when the given dynamic object is loaded, other systems call it something else, use a different prototype or may not call any function when the object is loaded. | 21:41 |
EHeM | Unix's handling of libraries is less than wonderful (most other OSes you do a LoadLibrary() or such, then GetSymbol(), whereas Unix most libraries are statically linked as the last build step). | 21:43 |
_abc_ | I see more and more google sabotaging firefox esr as used in devuan. Anyone else seeing this too? I have 60.9.0esr and I have problems with captcha all the time lately. Even yahoo uses google captcha now. | 22:18 |
specing | perhaps you should reduce your reliance on sites relying on google | 22:21 |
specing | and google probably sabotages all firefox versions | 22:21 |
_abc_ | I do not rely on google at all but they get to use google more and more | 22:21 |
specing | they have their own browser to prioritise | 22:21 |
_abc_ | Indeed. And that's more than slightly illegal. | 22:22 |
_abc_ | Especially in Europe afaik. | 22:22 |
_abc_ | https://groups.yahoo.com/neo/groups/LTspiceFiles/info example. I have a yahoo id because I am a member of the parent group LTSpice, but I can't even see the captcha being created when I try to join that subgroup. | 22:23 |
_abc_ | It says "Captcha Creation Failed" in red font. | 22:23 |
_abc_ | I just checked I have the highest version of ff esr available in ascii | 22:23 |
EHeM | _abc_: I'm not certain it is deliberate sabotage, so much as wanting to use the latest and greatest features in Chrome, but aren't yet in Firefox or don't even have a written standard. | 22:23 |
golinux | I have not seen a captcha in a very long time. | 22:24 |
_abc_ | Strange, they are everywhere. | 22:25 |
EHeM | golinux: You don't have any web browser on your system, merely NNTP and an e-mail client? | 22:25 |
_abc_ | The new "do nothing" captcha is particularly obnoxious, it uses google access records to qualify the user as human. No google access, not human. Really? | 22:25 |
_abc_ | I use duck and google and a few other things alternately and all with js and tracking off. | 22:26 |
specing | the new "do nothing" captcha is using human labour to train their waymo AI | 22:26 |
EHeM | Often I use `Lynx` when in need of a search engine. | 22:26 |
_abc_ | As a result I assume my "footprint" does not exist unless they collate browser data across sessions. | 22:26 |
_abc_ | specing: really? How the heck? | 22:27 |
specing | something needs to be done about cookies | 22:27 |
specing | is there a "noscript" for cookies? | 22:27 |
EHeM | `ln -s /dev/null cookies.txt` | 22:27 |
specing | _abc_: easy, you check "I am human", then you get to pick tiles with street signs in them | 22:27 |
_abc_ | I don't even get that far. | 22:27 |
EHeM | /dev/null can store *millions* of cookies. | 22:28 |
_abc_ | I have clean cookies at session end enabled. | 22:28 |
specing | _abc_: do you close ff very often? | 22:29 |
specing | > And that's more than slightly illegal. | 22:30 |
specing | Takes years to prove in court and the fines are pennies | 22:30 |
golinux | EHeM: Chromium, FF and palemoon | 22:46 |
avbox | Does someone know how to switch from LineIn to Speaker in pulseaudio via command line? | 22:47 |
_abc_ | specing: In Europe it's sort of built in that such things are illegal. | 22:47 |
_abc_ | specing: the equivalent of a US class action suit, massive, on behalf of about 300 million people at once. | 22:47 |
_abc_ | Even if each gets $10 payout they'll think twice before trying again. | 22:48 |
_abc_ | And yes I close ff at least several times per day. Sometimes deliberately. | 22:49 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!