#ossasepia | 2019-10-12

diana_coman: it is done!! i have reproduced fully the dulap-baking process. will deedbot the new, corrected and tested recipe, shortly.

03:50

diana_coman: http://logs.ossasepia.com/log/trilema/2019-10-11#1945002 .

03:50

(trilema) 2019-10-11 asciilifeform: !!deed http://www.loper-os.org/pub/dulap_construction_kit_r2.txt

03:50

diana_coman: would not have taken so long as it did, but this is VERY slow box. kernel compile 2+hrs.

03:53

the test machine booted as-expected on 1st try, and is running now.

03:55

diana_coman: i did not use 'apu1' , as it req's a serial port console to be config'd, and not applicable for your use case. but in the coming days i'ma replicate on a 'apu1' for own use.

03:58

diana_coman: the basic process, nao 100% proven to work as printed in the new recipe. if it so happens that your irons need kern config changes, i'ma help to walk through this tomorrow.

03:58

*

asciilifeform with working box, nao, finally to bed.

~ 3 hours 55 minutes ~

07:54

http://logs.ossasepia.com/log/ossasepia/2019-10-11#1005672 - *when* should you have asked this?

07:54

Logged on 2019-10-11 16:59:36 whaack: a little update on my interests post. first, i thought this was understood but i want to clarify with you diana_coman that this post is not supposed to be a continuation of previous/future tmsr work but instead a list of the things that i am personally interested in. (that would be aeparate post that will come later.) second, i've come up with six categories: pr

07:57

whaack: happily for you this time, the post is indeed a list of your interests, not a revision of the previous post, no; outline at this stage sounds ok but it's hard to say much on it at that level (other than you seem perhaps to overwork the classification before you even got in clear exactly what content you have; it generally works the other way around)

07:57

asciilifeform: thank you for the updated recipe, I'll try it out today.

~ 50 minutes ~

08:47

asciilifeform: done, booted successfully, I'll have to still check everything but will run it also on at least one more amd today

08:52

and I tweaked an option at kernel compile time so my usual kbd works as well, lolz.

08:53

the sort of thing that works way better when not totally tired like yest.

~ 1 hours 1 minutes ~

09:54

asciilifeform: hm, apparently the recipe though results in / mounted as read-only ?!

09:58

from dmesg: http://paste.deedbot.org/?id=45Rh

09:59

I don't get why it's trying to mount it as ext3; fstab clearly say ext4 for /dev/sda2, ugh

10:09

mk, I added to boot line rootfstype=ext4 and that at least got read of trying to mount it as ext3; onth, it's STILL read-only, grrr

~ 21 minutes ~

10:30

got rid*

10:30

at least remount works fine so hm

~ 2 hours 6 minutes ~

12:37

guten tag diana_coman . lemme check re these oddities, 1s

12:39

diana_coman: confirmed, a) root mounts ro . will find out why ( did not occur, naturally, on actual dulap ) b) sad python. confirmed. may have occurred on actual dulap and fixed after imaging (or phuctor, lol, could never have worked ) . didja ever look on smg-primary ? had to also fix there ?

12:39

morning asciilifeform ; let me know what you find; I'll be around

12:40

asciilifeform: I don't recall having to fix python, no

12:40

then again, eulora doesn't actually use any python as such

12:41

meanwhile it's on my to-check list, hence ?!

12:41

i suppose we see when you get the backup cut up

12:41

mine, naturally, was troo-python (none of my proggies will execute under 3, it's exotically broken)

12:41

asciilifeform: the idea is/was that if I'm fine with the backups I have, then disks get destroyed, not sent about

12:41

diana_coman: i rec to recreate machine before burnin'

12:42

and my backups are of smg data, not of the system

12:42

aha

12:42

asciilifeform: this is why I'm installing dulaps, no?

12:42

diana_coman: righto

12:42

diana_coman: fwiw i had my disks sent, cuz they did not contain any proprietaries that hadn't already been stolen w/ the last dulap

12:43

anyway i'ma go vivisect the tester, brb

12:43

asciilifeform: well, eulora server is a bit of a different thing, hence rather not send unless have-to

12:43

kk

12:44

I'll be away for about 30 minutes and then back at console

12:45

np

12:45

i expect will take at least that long, still waking up as the sun goes up.

12:54

diana_coman: i expect (1) is sumthing to do with ext4, on orig dulap had set up w/ reiser but did not want to propagandize the use of obsolete fs . looking into why mounts ro.

13:07

btw i also used mbr, rather than gpt, on orig, updated recipe so people could use >2tb disks. may be culprit (req's diff options)

13:18

asciilifeform: I was rather surprised at using gpt tbh; and why do you start first partition at 64?

13:19

diana_coman: was looking just nao re why. apparently no good reason, carried over from rk recipe (where the chipset actually required it) , can just as well start from 1

13:20

diana_coman: also found culprit re why mounts readonly ext4. the boot stick mkfs.ext4 enables 'huge files' opt by default, which needs kernel flag, aboutta rebuild w/ same

13:20

will add all of this to recipe.

13:20

kk, thanks

13:21

re python, apparently was in fact defect in the orig install, and had fixed immediately when setting up phuctor.

13:21

will add also to recipe. (or, alternatively, will make a new fs tarball)

13:23

still reading dox re (1) , cuz seems as if it may be red herring (applies only on 32bit kernels??)

13:25

diana_coman: re gpt -- >2tb raid arrays require it. i didnt have one, but intend to when rebuild dulap, hence the patch.

13:26

ah, ok re gpt.

13:32

update : confirmed that 'huge files' item is red herring, it is unneeded on 64bit kern. so continuing dig.

13:39

whaack

diana_coman: good morning. I should have clarified the task assignment upon receiving it. The reworking of my classification came after writing content and realizing that my 'programming' section was ~identical to my cryptography section. This follows from the fact that work related to crytographic tools is the ~only interesting programming work. At least for me.

13:42

whaack: morning; programming ~= controlling computers and if you want to do crypto, you kind of need to actually control your hardware so unless you have something other than existing computers, you need to know stuff around crypto itself as well; that being said, there's no problem in saying that your interest is in crypto really and that's it; ie you will learn anything else you need for & around

13:43

it but that's your goal, sure.

~ 34 minutes ~

14:17

diana_coman: found the eggog. without ancient notes, would never have found, troo cascade of liquishit. so, when made this tar, used 'exclude' command, with 'root'. and wrong regexp. so in fact 'evaporated' the initd script in /etc/init.d/root . which is what remounts / as rw .

14:19

diana_coman: i suspect that the Right Thing nao is to roll new tarball , replace the prev.

14:19

but 1st lemme demostrate the fix :

14:20

1st remount as rw : mount -o remount,rw ; then this goes in /etc/init.d/root ; then rc-update add root boot ; then reset. get proper boot.

14:21

incidentally in /etc/fstab, prolly want ramdisk,

14:21

shm /dev/shm tmpfs nodev,nosuid,noexec 0 0

14:24

i'ma make new dump tarball.

14:24

incl. correctly-sel'd python.

14:24

asciilifeform: all right; what's the eta re new tarball?

14:25

hr+. gotta write to slow usbstick and then upload to snail server.

14:25

I did precisely the remount earlier but anyway, didn't fix python otherwise so didn't really want to go ahead with all the rest on it as it is

14:26

fwiw at least the recipe worked with same results on a 2nd amd on a gigabyte board

14:26

diana_coman: i'ma start.

14:26

so there is some support to "will work on various amd platforms"

14:26

asciilifeform: ok, thanks.

14:26

diana_coman: it's expected to work on any amd after k7.

14:37

1s, gotta redo, forgot to 'make clean' in '/usr/src/linux', will massively bloat the tarball

14:44

heh

14:47

uploading.

14:50

steps 3-12, 3-18 should no longer be needed.

14:52

upload oughta be done in ~30m

14:53

sha512==51b339fd922497bbd11abc38375b54beb01a9737af55dd4114c328023927ce07e2bd90a6c9b1a3f793b862dc951b0ae438bfd3591c295254982c880c89190787 meanwhile.

14:53

diana_coman: http://thimbron.com/?p=282

14:54

diana_coman: if you have a ftp going on 1 of those cheapo leased machines, i can try and put there, faster.

15:03

thimbronion: answered.

15:05

asciilifeform: I didn't set up any, uhm.

15:05

diana_coman: 45% in

15:05

still goin'.

15:06

so prolly it will take ~same anyway

15:06

this bugger is slow but not as slow as the bmore monkeys, last word from whom was 'i'ma get your order to the boss' who apparently took vacay

15:09

diana_coman: while wait : when you do the kernel build, plox to examine whether it has the cp2101 (FG) thing , and the flags for iptables

15:10

thimbronion: was my own text on the role of feelings clearer/easier for you than hanbot's?

15:10

Logged on 2019-10-05 11:03:54 diana_coman: thimbronion: re your question on http://thewhet.net/2013/09/your-feelings-are-out-to-get-you/comment-page-1/#comment-55392 - you are confusing the contexts there (or not really making a proper separation at all); read http://ossasepia.com/2017/01/25/feelings-are-helpful-but-not-for-idiots/ and see if that answers it for you; if it doesn't, ask me again.

15:10

asciilifeform: I thought iptables was deemed utterly un-necessary?

15:10

diana_coman: iirc you used it

15:11

i -- left off

15:11

well, not after the discussion and conclusion, no? otherwise what use is to discuss and conclude, lol

15:11

so no, not using anymore either

15:11

aite.

15:11

if anything, it would help to check that an FG works well

15:11

iirc there was the need to downgrade via companion or some such shit

15:11

point being that kernels are sorta intimate affair, it's like putting on another person's eyeglasses, to use kernel conf unexamined.

15:12

(it's in my notes somewhere, will get to it when reproducing all stuff anyway)

15:12

CONFIG_USB_SERIAL_OPTION

15:12

asciilifeform: I don't think this is very sane though for non-personal machines; precisely because if every machine has to have different kernel conf than it's not all that reproducible

15:13

sure, on my personal machine, I fuck the kernel whatever way I want; not same for non-personal machines though.

15:13

diana_coman: the issue is that erryone has slightly diff machine. esp if they aint all built by 1.

15:13

for instance if you have a raid, you gotta have the flags for ~that~ raid.

15:13

myeah

15:14

but I'm trying to keep those differences to a very minimum, ideally 0, ugh.

15:14

the offered kernel is made so as to guarantee bootability on commonplace amd irons. but it cannot be guaranteed to operate ~arbitrary~ irons w/out some tailoring.

15:15

this is consequence of 'tower of babel' of pc irons.

15:15

myeah, I know.

15:16

fwiw it stood up on a circa-2010 box last night w/out changes.

15:17

eta <20m

15:17

i'ma go wash up , brb

15:18

crowncloud: are you there?

~ 16 minutes ~

15:34

diana_coman: yes it was.

15:38

diana_coman: http://www.loper-os.org/pub/dulap-r2.tar.gz . checksum pasted earlier.

15:39

diana_coman: when this item is confirmed to pass basic smoke test, i'ma fix the recipe & sign.

15:39

( unless diana_coman would rather post own recipe )

15:39

tarball will sign in any event.

15:45

diana_coman: lemme know as soon as you get chance to try , cuz uploading these to my 1 remaining box takes ~45min. ea. time.

15:47

asciilifeform: ok, downloading it

15:47

asciilifeform: why is that box so slow, I don't recall?

15:47

diana_coman: it's where my www. kolhoz .

15:48

at any rate, if that's the problem, I'll set up an account for you on younghands or somewhere, lolz

15:48

diana_coman: at the rate bmore is going, i might need another, so ty.

15:48

'elderlyhands'

15:48

asciilifeform: do you still plan to send a box to maxim/mivo?

15:49

heh, quite

15:49

diana_coman: indeed. already purchased the disk for that 1

15:49

kk

15:49

i expect it'll take time to go across atlantic.

15:49

diana_coman: i'ma colo a 'apu1' with him. will see how goes.

15:50

probably, yes; though not necessarily longer than bmore takes to give you a box locally from what I understand

15:50

entirely possib

15:50

about 500M downloaded

15:50

ha, much faster than when up

15:50

they must have an asym. pipe.

15:51

usually for kolhoz sort of thing they do as far as I saw

15:51

asciilifeform: iirc you can get that with him via email only ie without account on website, correct?

15:51

ftr i asked the bmore people about ddos. after ~that~ was told 'gotta ask boss' and made to wait.

15:52

diana_coman: i haven't opened any type of acct at maxim's just yet, still in conveyor.

15:52

asciilifeform: heh; I asked some some "eu" thing; first they replied with "oh no, it's nothing to worry, as long as you update/blabla"; after which, when I asked specifically and exactly what do they do if there's an attack on MY box...silence

15:52

I don't even expect they'll come back to me, lolz

15:53

asciilifeform: try maybe you can get away without account; worth a try at least and as far as I can tell, if there's a chance of that, so far it's with them only.

15:53

diana_coman: ah, but see, the bmore people bill for packet mass. so 1st gotta get answer to 'what when ddos?' cuz on my surface reading, the answer may be 'potentially infinite bill' and then no go

15:54

asciilifeform: heh, the eu had "potentially disconnect you", hence my q ~= wtf do you think I'm paying you for to disconnect me??

15:54

anyways, supposedly one way or another I'll get my blog back up and then do the write up of all that shit

15:54

diana_coman: after latech i aint sure whether anyone can be believed, re 'when ddos'.

15:55

only way is recon by fire.

15:55

asciilifeform: at the very least they *have to say it in writing*

15:55

that's the rub

15:55

and I fully intend to rub them in it

15:55

BingoBoingo is busy rubbin' atm, iirc.

15:55

i dun expect much to come from it.

15:56

asciilifeform: tarball downloaded and checksum checked, all fine

15:56

ok, then ready for testfire.

15:56

asciilifeform: no, the point is *before* anything; because they won't quite want to say/admit "yes, you pay for us to disconnect you" but onth if they say "no, we don't", they are in a bigger pickle when they do so...

15:57

hence why I expect they won't come back to me at all, no

15:57

diana_coman: which firm was this ?

15:57

eu i mean

15:58

asciilifeform: some eunet iirc or if you really need it I'll look it up in the notes; a quite big corporation style thing but at that point I tried them too, why not

15:58

aok

15:59

asciilifeform: are you writing the new recipe now?

15:59

i am. the only diff tho is http://logs.ossasepia.com/log/ossasepia/2019-10-12#1005754 + the new tarball url atm.

15:59

Logged on 2019-10-12 10:50:45 asciilifeform: steps 3-12, 3-18 should no longer be needed.

15:59

ah, then I'll try to proceed with it

15:59

ty diana_coman

16:04

diana_coman: http://www.loper-os.org/pub/dulap_construction_kit_r3.txt and will deedbot after verified to work

16:05

asciilifeform: kk, thanks; working on it

16:06

asciilifeform: you should update that parted to start at 0 too

16:06

a yes

16:07

diana_coman: i'ma hold off on rewrite of instrs. until you return 'it worx'

16:07

may need other fix.

16:08

kk

16:08

parts cant start at 0 btw, that there's the mbr

16:09

lilo needs to live somewhere too

16:09

so usually people give it coupla sectors

16:09

yeah, 34

16:10

thimbronion: is the review-trouble clarified now?

16:12

asciilifeform: you forgot to update the name of the tarball in the new recipe at 3-16

16:12

3-11, ugh

16:13

indeed.

16:13

*

asciilifeform will put these in as diana_coman emits'em

16:14

diana_coman: btw i found that if using 'grub' you actually needs those 63 spare blox.

16:14

looked on a local box w/ 'grub', partitioned same.

16:14

'tis a pittance tho, block==512b.

16:14

diana_coman: yes but I am absolutely certain I will run into more questions when I try again tonight. Got to head out now.

16:15

asciilifeform: now the tar complained about timestamp; do you know if that is more than just a warning?

16:15

I never saw this before

16:16

diana_coman: our clocks differ, i suspect

16:16

so yours thinks 'from future'

16:16

asciilifeform: possibly; my q is whether it still unpacked everything or not

16:16

these get tar'd with 'p' flag, i.e. preserve-all incl. times

16:16

diana_coman: oughta have

16:16

seems so as far as I can see; ok, moving on

16:17

looked, box where tar'd set to utc, so yes 'from future, martians'

16:17

it's oct 13 on it.

16:17

( this problem, will fix self )

16:25

thimbronion: ok but you know, ack still needed on positive outcome too; don't do this "no news is good news"

16:26

compiling kernel, changed it to armoire at least to make sure it's the one I'm aiming; will remember to change in lilo.conf too, lolz.

16:26

diana_coman: in the end may want to make yet another tarball, it is quite bare, and lacks such amenities as e.g. gnat.

16:27

asciilifeform: that I have & will deploy afterwards anyway

16:27

aite

16:27

speaking from the pov of hypothetical 'make 6 spare boxen' and similar.

16:28

asciilifeform: I suspect I'll have to make scripts really

16:28

diana_coman: well trinque already made scripts. but iirc you had an item that still not 100% certified to live in cuntoo.

16:28

asciilifeform: is that rootfstype=ext4 parameter needed in the end?

16:28

diana_coman: per the dox, seems to be. hence wai put in.

16:28

I had added up on previous install to get rid of idiotic attempt to mount it as ext3 (and it worked) but dunno if still needed it now

16:28

ah, ok

16:29

i dun use ext4 myself, so had to look up.

16:29

asciilifeform: yes sadly, can't yet move on to cuntoo.

16:29

right, is what i recalled.

16:29

plus honestly, cuntoo is still work in progress in fact, not yet really there.

16:29

more like "work in waiting" but anyway

16:29

( if could use, diana_coman would not have asked for this earlier barbarism, i inferred also )

16:31

ftr 'dulap-gentoo' is, concretely, the last gentoo built by asciilifeform while traditional gentoo still was buildable ; incorporating the ban flags & cleanups discussed in '15-'17.

16:31

asciilifeform: certainly I'd have used cuntoo if I could.

16:32

python3, ftr, was not entered into the ban list strictly because asciilifeform uses, sadly, 1 single program (ice40 toolchain) that runs in it. otherwise can be safely discarded afaik.

16:34

asciilifeform: success! booted Armoire-II; / is mounted in rw mode, gcc is 4.9, python is 2.7

16:34

a+++ diana_coman

16:35

diana_coman: didja end up fiddling w/ kern. conf ?

16:35

thank you for the new tarball asciilifeform !

16:35

np diana_coman

16:35

I looked about, changed the one for my usual kbd (the emergency ps kbd is not that comfy on fingers) and otherwise looked around but I didn't see something directly needed/to change

16:36

diana_coman: lemme know when you have chance to mirror the tar, mine is veheheheryslow

16:36

would rather link folx to mirror.

16:36

diana_coman: lemme know if you want other changes in the doc, i'ma revise , sign , deed.

16:36

I'm sure it can be tuned probably but I'm not all that interested in that + atm I really need to get on with installing all the stuff and bringing the eulora servers up at least here

16:36

diana_coman: re kernel -- may have to turn knob to get FG.

16:36

depending on what cables you have.

16:37

asciilifeform: hm, in this one it's on usb; do you know where is that knob in kernell?

16:37

*

asciilifeform looks

16:38

i had to CONFIG_USB_SERIAL_PL2303 for mine

16:38

if you have the cp210x one ought work as given in tarball tho.

16:38

I recall that is the knob (or at least one of the needed ones), yes

16:38

no, should be pl2303, ugh

16:39

most people got the cp210x

16:39

hm, need to look in notes and check

16:39

what colour is your plug ?

16:40

asciilifeform: I have both black and blue (usb 3)

16:40

then you need the knob

16:40

blue's a pl230x

16:41

hm, and black isn't ?

16:41

black was the chinesium, it actually came in the 2 variants unlabeled, annoyingly

16:42

initially i raged about it, then thought 'good to have mix, will see if there's oddities/noncomplicances in standard'

16:43

my criteria for these was 'anyone but ftdi'

16:44

diana_coman: looking again at my notes, turns out that no, all batch2 fg people have the pl.

16:44

asciilifeform: this is what I recalled, that it was ~always pl

16:44

so yes, knob.

16:45

so I think at least mention the location of that knob, in the recipe perhaps

16:45

aha, you should have 2 pl. (regardless of colour)

16:45

diana_coman: will put in.

16:45

*

diana_coman goes back to recompile

16:47

diana_coman: dun fughet this if you're plugging in 3 or moar.

16:47

ericbot

(trilema) 2018-08-01 asciilifeform: dwc_otg.speed=1 on kernel param line.

16:50

asciilifeform: good point; add to recipe too, please

16:50

will.

16:53

the bit re kbd is interesting -- none of the boxes where i had this going, ever had a classic ps/2 kbd.

16:53

tried even last night several usb kbd, all worked.

16:54

*

asciilifeform brb, eating (very late breakfast lol)

17:06

asciilifeform: either I'm blind atm or it's one of those that depends on something else and so on ffs; I went to Device Drivers -> USB support but I can't find it anywhere, grr

17:11

ah, found it; it's in device drivers/usb support/usb serial converter support/usb prolific shit

17:11

gah

17:12

I will need to go for a few hours so I'll let it compile and then get back to it when I'm back

17:17

*

asciilifeform back

17:30

asciilifeform: confirmed compiled, booted and FG seen; now I'm going away for 2-3 hours, will move on to rest of installs after that

17:30

a++

17:30

diana_coman: lemme know if again puzzler, i'ma also walk for some hrs.

17:31

will check log later today.

~ 1 hours 46 minutes ~

19:18

diana_coman: noted re: good news.

~ 1 hours 12 minutes ~

20:30

I'm trying to write ebuild for Eulora client, but will alo require separate ones for cal3d/crystalspace it appears.

20:31

shinohai: yes, cal3d and crystalspace are the ones to start with because they will stay the same, most likely.

20:31

cal3d should be the easiest to start with really

20:32

for one thing getting it to build with gprbuild is quite straightforward and for the other, it's way less messy than cs

20:32

shinohai: so I'd say start with that and let me know how it goes

20:32

Are they simply a specific version, could i mask them if already in portage?

20:33

shinohai: they are a specific version; I doubt they are in portage but look and let me know, why not

20:33

http://logs.ossasepia.com/log/ossasepia/2019-10-12#1005952 <<< ebuild has hard requirement for gcc 4.9.4 / Ada

20:33

Logged on 2019-10-12 16:32:21 diana_coman: for one thing getting it to build with gprbuild is quite straightforward and for the other, it's way less messy than cs

20:33

you'll need the exact versions used/mirrored by s.mg though, at least for the client as far as it'll be released (other than that you can always do your own /make it work with any version, ofc)

20:34

Thats why I think I should just use the ones I have mirrored on my site, because exact match for ones from minigame

20:34

shinohai: what do you mean? they certainly compile with gcc 4.9.4 ofc; and fwiw I compiled cal3d with gnat/gprbuild, it worked fine

20:34

(So ebuild is better method of doing this, I believe)

20:34

cs is messier because it has a ton of if-defism among other warts

20:35

shinohai: so yes, sure; use those; not sure in the end what are you trying to do exactly? (I thought you were trying to *make* ebuilds for those versions, yes)

20:36

Yes, that's what I intend to do (write ebuild for all 3 components), just assuring I couldn't be lazy and just do package.mask xD

20:36

note that cs also has quite a few dependencies so you'll probably be better off doing first cal3d and then starting from those cs deps and only at the end the cs itself

20:37

shinohai: first of all: are they even in portage? because I seriously doubt but just check anyway, it's not a difficult thing to do

20:37

yeah I checked, both available

20:37

if nothing else, an existing ebuild might help you as an example perhaps

20:38

ah; what versions compared to minigame ones?

20:38

cal3d is quite frozen so maybe lucky there

20:38

emerge search shows:

20:38

media-libs/cal3d

20:38

Latest version available: 0.11.0-r1

20:38

Latest version installed: [ Not Installed ]

20:39

the 0.11 sounds right, dunno about r1

20:39

so that might be handy if it works, worth a try at least

20:40

Will report in when I have results o7

20:40

as I said: cal3d has been quite frozen for ages, so perhaps the easy way works for it

20:40

shinohai: ok, good luck

~ 45 minutes ~

21:25

asciilifeform: mysql seems to try to pull in openssl (some of its deps) despite me specifying even same version of mysql I had hammered to work on the test server, ugh; this will be some fun, by the looks of it

~ 30 minutes ~

21:56

*

asciilifeform waves to diana_coman

21:57

so, lessee :

21:57

diana_coman: do you remember how this gentoo was made ? in /etc/portage/package.mask/crapolade : there are banned pgkgs ;

21:58

diana_coman: in /etc/portage/make.conf there are 'minused' gentooflags .

21:58

diana_coman: ssl was not in the scope of either, if you had a custom ban, it was 100% your own work

21:59

tho if you dun recall how you did it, dollars to doughnuts it was by putting a >=openssl-bannedverwhatever in 'crapolade' .

21:59

asciilifeform: I banned it specifically because otherwise it gets pulled in and conflicts with libressl

21:59

right, so this was in your working piece but not in mine

21:59

asciilifeform: actually that is what I should do

22:00

it's ~99% of the win from using gentoos, the fact that you can do this

22:00

ie so far I specifically masked it for dev-db/mysql but I should add it to crapolade

22:00

so it *did* help I complained here !

22:00

crapolade perma-bans

22:00

diana_coman: dun hesitate to complain even if re general operation of dulapgentoo and not concretely the given tar

22:00

i might know the answ.

22:00

thanks!

22:01

np.

22:01

there are other packages you might haveto ban depending on what's being built. i cannot claim to have mapped out the entire known universe of liquishit.

22:02

my orig ban concerned concretely poettering et al and their 'contributions'.

22:02

asciilifeform: sure; I *do* have my notes, but apparently at that time it was enough to ban it for mysql specifically ; possibly because other versions of the deps already installed or something (ie it's pulled in by a dep of mysql and it has MANY, ugh)

22:03

fortunately i also -- have notes. but i used quite diff set of proggies , from smg's.

22:03

aha

22:03

I'll get to it, lots of stuff to compile and check still, ugh.

22:03

re mysql, iirc BingoBoingo set up, worked, on rk.

22:03

(rk is ~= this gentoo, but for arm64)

22:04

asciilifeform: well, I set *this versione exactly* on the test server so it has to work in the end.

22:04

for that, put that ver in package.use

22:04

BingoBoingo

asciilifeform: Yes, just for a bloggatron on RockChip the secret MySQL sauce is some config (rather than compile) changes that keep it from eating all the ram.

22:04

diana_coman: this forces concrete package.

22:05

I've just masked anything with higher version really

22:05

the opposite of package.mask.

22:05

that also worx.

22:05

( when you mask, it uses the latest permissible ver. )

22:10

BingoBoingo: plz see if you can excavate the magick flag for mysql

22:13

*

asciilifeform tired and will go horizontal but will check here again before proper bed; bbl

22:13

asciilifeform: ah, shit; openssl is *installed* actually?

22:13

BingoBoingo

default-storage-engine = MyISAM is the MYSQL config flag to productively constrain MYSQL ram use on limited ram

22:13

BingoBoingo

https://archive.is/7Vfxd << Lobbes guide to mysql on rockchip

22:13

diana_coman: installed, nginx pulled in

22:13

and curl

22:14

uhm, curl is not installed and precisely, I was setting up to install curl *without openssl"

22:14

*

asciilifeform genuinely bbl

22:14

but uhm, I thought the idea was no openssl really

22:15

diana_coman: gentoo pulls tarballs from portage repos, and afaik cannot be installed w/out ~some~ sslism

22:15

i have not experimented with alt-ssl's at the time of the writing of dulap-g.

22:15

hm; so only cuntoo is ssl-free?

22:15

afaik

22:15

hm

22:16

diana_coman: in principle possible to ban the pkg and then remake world. but i have not attempted.

22:16

none of my proggies used sslism for anyffin. (i did not have trb on this dulap, and trb carries own ssl tarball)

22:17

given that I should have the servers up by tomorrow evening, I'd rather not rebuild world, hm

22:17

so I suppose I'll have to live with ssl for now at least.

22:18

diana_coman: i expect you'll have >1 of these inexpensive boxen, from this can win ability to make constricted sets w/ rebuilt worlds etc. when time permits.

22:19

*

asciilifeform genuinely must bbl ; diana_coman et al plox to put in any q's re this item, in log, will answr. when i get back.

22:22

gn asciilifeform

~ 51 minutes ~

23:14

fwiw going through the old notes re install on 1st smg server @pizarro (aka the production server), it turns out that indeed, I had to eselect python 2 so as to have python 2.7 as default.

~ 30 minutes ~

23:44

diana_coman: how much time should I be spending on this review?

23:49

I'm at about 3 hours, but I'm not happy with it, and I'm running out of time.

23:58

thimbronion: that's the sort of question that you can answer best; what's wrong with it/why "not happy"?

23:59