WTF Bites
-
@Atazhaia Sundbyberg, men som konsultslusk så drar jag runt en del.
Also for you merkins: Börkbörkbörk.
-
pjh@hpdesktop.com $ git clone git://git.openssl.org/openssl.git
Cloning into 'openssl'...
remote: warning: unable to access '/root/.config/git/attributes': Permission denied
remote: Counting objects: 321124, done.
remote: Compressing objects: 100% (87666/87666), done.
remote: Total 321124 (delta 229535), reused 315400 (delta 224036)
Receiving objects: 100% (321124/321124), 66.79 MiB | 8.13 MiB/s, done.
Resolving deltas: 100% (229535/229535), done.
pjh@hpdesktop.com $
-
@Carnage Ah, Fjollträsk. Men de första 4 bokstäverna stämmer överens iaf.
-
PS R:\> git clone --depth=1 git://git.openssl.org/openssl.git Cloning into 'openssl'... remote: warning: unable to access '/root/.config/git/attributes': Permission denied remote: Counting objects: 18313, done. remote: Compressing objects: 100% (15203/15203), done. remote: Total 18313 (delta 884), reused 16714 (delta 579) Receiving objects: 100% (18313/18313), 11.29 MiB | 559.00 KiB/s, done. Resolving deltas: 100% (884/884), done. Checking out files: 100% (18181/18181), done. PS R:\> ls .\openssl Directory: R:\openssl Mode LastWriteTime Length Name ---- ------------- ------ ---- d---- 21-Nov-18 01:36 PM .github d---- 21-Nov-18 01:36 PM apps d---- 21-Nov-18 01:36 PM boringssl d---- 21-Nov-18 01:36 PM Configurations d---- 21-Nov-18 01:36 PM crypto d---- 21-Nov-18 01:36 PM demos ...etc.
Checks out. ()
-
-
Today's showing that VS developers must be mods here ():
1>foo.obj : error LNK2022: metadata operation failed (801311E4) : Duplicate managed types have different visibilities. 1>LINK : fatal error LNK1255: link failed because of metadata errors
Which types? Which visibility for each? Who knows? Who cares? Why would you expect a compiler that has just detected a problem to actually tell you where it found the problem?
Something broke. Somewhere. Go figure.
-
@Gąska If you have git, clone git://git.openssl.org/openssl.git and see for yourself. ¯\_(ツ)_/¯
(yes, the directory is there. At least it was when I tried)
-
Today's showing that VS developers must be mods here ():
1>foo.obj : error LNK2022: metadata operation failed (801311E4) : Duplicate managed types have different visibilities. 1>LINK : fatal error LNK1255: link failed because of metadata errors
Which types? Which visibility for each? Who knows? Who cares? Why would you expect a compiler that has just detected a problem to actually tell you where it found the problem?
Something broke. Somewhere. Go figure.
The error is in object file. It should be obvious the solution is to remove that object file and let the program recompile. I mean, even if you knew which type exactly caused the problem, the only possible fix would still be removing that object file and recompiling.
-
@Gąska If you have git, clone git://git.openssl.org/openssl.git and see for yourself. ¯\_(ツ)_/¯
Looking at those root access errors, it doesn't look like perfectly safe operation. And I don't feel like spinning up a new VM today.
-
Which types? Which visibility for each? Who knows? Who cares? Why would you expect a compiler that has just detected a problem to actually tell you where it found the problem?
I'm getting overtones of Xcode here...
-
boringssl
Is this real?
It's so boring, it's empty.
$ ls -la openssl/boringssl/ | anonymise total 8 drwxrwxr-x 2 pjh pjh 4096 Nov 21 12:39 . drwxrwxr-x 23 pjh pjh 4096 Nov 21 12:39 ..
-
@PJH empty folder in Git!? Last time I checked, it was literally impossible by design.
-
@PJH empty folder in Git!? Last time I checked, it was literally impossible by design.
Pass.
boringssl
history > https://git.openssl.org/?p=openssl.git;a=history;f=boringssl;hb=HEAD
-
History isn't shedding much light either:
pjh@hpdesktop.com:~/src/openssl$ git log -- boringssl commit 4b5f7e7555340db28b285452cfd0cafe1854de2a Author: Matt Caswell <matt@openssl.org> Date: Tue Mar 14 10:18:21 2017 +0000 Update ossl_config.json for later BoringSSL commit Update the list of suppressions so that we can run a later BoringSSL set of tests. This also adds an ErrorMap to greatly reduce the number of failing tests. The remaining tests that still fail are just disabled for now. Reviewed-by: Richard Levitte <levitte@openssl.org> (Merged from https://github.com/openssl/openssl/pull/2930) commit ab29eca645cdb38ffe73d141bbd7c6879b602860 Author: Emilia Kasper <emilia@openssl.org> Date: Wed Nov 9 17:25:35 2016 +0100 Run BoringSSL tests on Travis Reviewed-by: Richard Levitte <levitte@openssl.org>
-
@Carnage Ah, Fjollträsk. Men de första 4 bokstäverna stämmer överens iaf.
Smørrebrød, smørrebrød, røm pøm pøm pøm?
-
Today's showing that VS developers must be mods here ():
1>foo.obj : error LNK2022: metadata operation failed (801311E4) : Duplicate managed types have different visibilities. 1>LINK : fatal error LNK1255: link failed because of metadata errors
Which types? Which visibility for each? Who knows? Who cares? Why would you expect a compiler that has just detected a problem to actually tell you where it found the problem?
Something broke. Somewhere. Go figure.
The error is in object file. It should be obvious the solution is to remove that object file and let the program recompile. I mean, even if you knew which type exactly caused the problem, the only possible fix would still be removing that object file and recompiling.
Can't tell if serious or not... Are you expecting the same code to not generate the error simply because you've deleted the object file and rebuilt it?
I did many things including ultimately a full rebuild from scratch (I guess I haven't anything better to do, and also my laptop makes a nice heater). No better. Well at least the same code reliably produces the same error, which I guess is something.
Too bad I'm apparently the only person in the company to get the error. That, or all cow-orkers either never compile the application or have had this error forever and always just went meh and ignored it (the error is just in one lib and for some reason doesn't prevent the rest of the build from working fine, I guess the lib must be dynamically loaded rather than directly linked into the exe or something like that)...
Not sure which one is the most scary.
-
Today's showing that VS developers must be mods here ():
1>foo.obj : error LNK2022: metadata operation failed (801311E4) : Duplicate managed types have different visibilities. 1>LINK : fatal error LNK1255: link failed because of metadata errors
Which types? Which visibility for each? Who knows? Who cares? Why would you expect a compiler that has just detected a problem to actually tell you where it found the problem?
Something broke. Somewhere. Go figure.
The error is in object file. It should be obvious the solution is to remove that object file and let the program recompile. I mean, even if you knew which type exactly caused the problem, the only possible fix would still be removing that object file and recompiling.
Can't tell if serious or not... Are you expecting the same code to not generate the error simply because you've deleted the object file and rebuilt it?
Yes. Speaking from experience.
I did many things including ultimately a full rebuild from scratch (I guess I haven't anything better to do, and also my laptop makes a nice heater). No better.
Well... You're in some very deep shit then. Depending on how much time you have to spare, and how much you trust particular tools, I'd start reinstalling libraries, compilers, package managers, trying out fresh OS install...
-
Bullshit, I specifically closed all my windows to make sure it would be free.
I love the consistency of Windows Update.
Leave the computer alone with pretty much nothing running so it can update automatically: "Device was actively in use."
Sit at the computer activelygamingworking: "Now is a good time to forcefully reboot and install updates!": AhHa! <edits code>
+++ b/src/Update.cpp @@ -845,3 +845,3 @@ void CClass::OnUpdate() } - if (bUpdate) + if (!bUpdate) {
-
@Gąska I really don't like the sound of that... The good thing is, it's in a lib that I don't actually use in whatever I'm doing (so I don't care if it's broken in my sandbox), but that's awful.
The other thing is that my laptop is probably going to be upgraded to Windows 10 fairly soon, which will likely go through a full reinstall, which will probably bring my productivity to 0 for some time, but at least that specific issue won't be my top one (and might even be solved!).
-
Today's showing that VS developers must be mods here ():
1>foo.obj : error LNK2022: metadata operation failed (801311E4) : Duplicate managed types have different visibilities. 1>LINK : fatal error LNK1255: link failed because of metadata errors
Which types? Which visibility for each? Who knows? Who cares? Why would you expect a compiler that has just detected a problem to actually tell you where it found the problem?
Something broke. Somewhere. Go figure.
I just had to track down a template error. Used to be that the last line of all the errors at least pointed to the line of your code that caused it. This one didn't. Just lots of stl header errors. Well, fuck. (All I did was move some code from one file to another in the same project - ) Finally tracked it down to a local static function that also needed to be moved that provided less-than functionality for a
GUID
. That took way too much time...
-
And now this is shifting from a VS to a cow-orker . I mentioned the compilation error I got to someone who answered
"I've been away for a week and just came back. I noticed we get this error in the automated builds since about a week, I've asked the devs in charge to look at it."
Indeed, I noticed errors in the build logs (and emails) we get from the build farm since about a week. I didn't investigate since they were not coming from any of my modules. Apparently, the devs in charge of that also didn't. The emails are sent to a mailing list that includes all devs, so they necessarily have received them. And even if you ignore that list, when one of your commits breaks the farm you get direct emails to you (in addition to those to the mailing list) every half-hour or so, with a very clear message ("one of your commit broke the compilation", basically).
So the guy who did that got automated emails through the mailing list. He got about 50 emails per day about this since last week. And yet not only is the issue still here, but he hasn't even sent an email to the list to say that he's working on it (which I assume he therefore isn't).
Remember what I said about having the hubris to believe that I'm somewhat better than most of my fellow developers? Yeah, sometimes it's definitely not hubris.
-
@Rhywden
Que?
-
-
@Gąska
HyperlooopSSL ?
-
@Luhmann with built-in tunneling support!
-
Status: Just noticed that viewing posts on a user's page, all the topics apparently have
RE:
prepended to them. Weird...
-
Ah yes. The less you buy, the cheaper it is option...
-
Ah yes. The less you buy, the cheaper it is option...
Naturally. There's only 5 left, so the more you buy, the rarer it becomes.
-
@Tsaukpaetra
Not all. Only those you have upvoted
-
@dcon The WTF there is that usomewhere in the product metadata the seller indicated that they're all 2.0 ounces.
My guess is that a "pack" actually means 6 of the individual 2.0 ounce size, which means that the "2-(Pack)" size should actually be 24 ounces (or $2.08/ounce), "3-(Pack)" should be 36 ounces (or $1.94/ounce), and so on.
Amazon sellers tend to be really bad about putting the wrong sizes in the product's metadata, so the $/ounce ends up being wrong. It's not Amazon's fault, though -- it's the sellers.
-
@anotherusername said in WTF Bites:
@dcon The WTF there is that usomewhere in the product metadata the seller indicated that they're all 2.0 ounces. My guess is that a "pack" actually means 6 of the individual 2.0 ounce size.
Amazon sellers tend to be really bad about putting the wrong sizes in the product's metadata, so the $/ounce ends up being wrong. It's not Amazon's fault, though -- it's the sellers.
Not when I clicked it. It showed 2 items.
-
@dcon It wouldn't be the first time that a seller put a misleading image of the product, either...
-
@anotherusername said in WTF Bites:
It's not Amazon's fault, though -- it's the sellers.
LOL: " Ships from and sold by Amazon.com. "
-
-
@anotherusername said in WTF Bites:
It's not Amazon's fault, though -- it's the sellers.
LOL: " Ships from and sold by Amazon.com. "
Still the sellers, IIRC. Via weird fullfillment magic.
-
boringssl
Is this real?
Basically: crypto software should not be exciting, it should be reliable and boring. (I.e. the opposite of CADT)
It’s a google fork
-
@Polygeekery said in WTF Bites:
I ask if they have rebooted recently, they say that they have.
Monitor was turned off, and then back on.
-
@Scarlet_Manuka said in WTF Bites:
@Polygeekery said in WTF Bites:
I ask if they have rebooted recently, they say that they have.
Monitor was turned off, and then back on.
Not quite as dumb as that. Hit the power button and it went to sleep, hit it again and it woke up.
Find the bigger WTF:
They put the machine to sleep and woke it up and thought that was a reboot.
That apparently they were trying to do a hard reboot, which, had it worked that way would have likely hosed the Windows install because Windows really does not react well to a hard reboot with updates pending completion.
-
@Polygeekery I think it's reasonable for an unsophisticated user to assume that "press the power button and everything goes off, press it again and everything comes back on" is in fact doing a reboot.
I also don't think they were necessarily trying to do a hard reboot - they probably don't even understand what that is. I expect they were trying to do a shutdown-startup cycle, and if the power button behaviour had been set to shutdown, it would have worked.
I personally would like to be able to update as part of shutdown. But these days 95% of the update work seems to be deferred until startup, so (as happened last night) when there are pending updates I do "update and restart", then shutdown; it's silly and wasteful of my time, but better than doing "update and shutdown" and then turning on the computer in the morning and being unable to use it for 45 minutes while it finishes updating (including five minutes of the clearly inaccurate "100% complete" message).
You could consider asking your users about restarting, rather than rebooting. It's a tiny linguistic change but does correspond with the terminology displayed by Windows, so may be more likely to trigger the correct course of action.
Or just change the power plans so that power button = shutdown, I guess.
-
If we had a sqlite thread I'd bump that but I assume we don't have an sqlite thread and ever if we do searching is stoopid, so here we go.
SQLLite is dumb. And I will always call it SQLLite, because the name they picked for it is dumb. sqlite. That's either SQL Ite, or SQ Lite. Neither is correct, and neither is this idiotic "database".
I had need to rename a column in a SQLLite database. So I did what anyone would do:
ALTER TABLE FuckYou ALTER COLUMN GiveMeMoney int
But nope. That doesn't work. Why? Because SQLLite doesn't support renaming columns. Because fuck you, that's why.
So how does the official SQLLite documentation suggest renaming a columns:
- Turn off PK checks
- Rename the TABLE to FuckYouOld
- Create the table FuckYou again, but with the column renamed
- Run INSERT INTO FuckYou (every fucking column) SELECT every fucking column FROM FuckYouOld
- Drop table FuckYouOld
- Turn PK checks back on.
And in case you missed it in that list-- they don't support renaming a column, but they support renaming AN ENTIRE FUCKING TABLE.
-
@Lorne-Kates said in WTF Bites:
And in case you missed it in that list-- they don't support renaming a column, but they support renaming AN ENTIRE FUCKING TABLE.
Yeah, kinda weird. I know I have to do that if changing the column type in MSSQL, but changing what should be metadata? Odd...
-
@Carnage Ah, Fjollträsk. Men de första 4 bokstäverna stämmer överens iaf.
Smørrebrød, smørrebrød, røm pøm pøm pøm?
Genau.
-
@Lorne-Kates said in WTF Bites:
I had need to rename a column in a SQLLite database. So I did what anyone would do: ALTER TABLE FuckYou ALTER COLUMN GiveMeMoney int
But nope. That doesn't work. Why?
Because that's not the correct syntax for renaming a column, in any database.
-
@Lorne-Kates If you don't care about column order you can just add a new column and copy the values over.
Btw, it seems like in newer versions you now can rename columns. Just downloaded a new one and tried it:
SQLite version 3.25.3 2018-11-05 20:37:38 Enter ".help" for usage hints. Connected to a transient in-memory database. Use ".open FILENAME" to reopen on a persistent database. sqlite> create table FuckYou (frist int, GiveMeMoney int, thrid int); sqlite> alter table FuckYou rename GiveMeMoney to GiveMeMoreMoney; sqlite> .schema CREATE TABLE FuckYou (frist int, GiveMeMoreMoney int, thrid int); sqlite>
-
@Scarlet_Manuka said in WTF Bites:
I personally would like to be able to update as part of shutdown.
Those of us who have to take our work laptops home at the end of the day like it at startup.
-
@Scarlet_Manuka said in WTF Bites:
to be able to
Not mandatory. Seems fine to me.
(although I've never had issues because I still use 7. Forced updates are usually unfun in any scenario.)
-
@kazitor Our corporate updater (Standard microsoft tools? What are you talking about?) likes always-on-top dialogs covering a quarter of the screen at the strike of noon. Options being 'update now' and 'update in 24 hours'. Supposedly it'll update if you next turn off the machine as well if you dismiss it but I've had that not work a couple times.
-
If you don't care about column order
Is there ever a case where you do care?
-
@Gąska Obsessive compulsion.
-
@Lorne-Kates said in WTF Bites:
SQL
Lite isdumbsimple.Yes, that's one of its primary selling points, and it's a bit mean to pick on it because of that. If you want an actual, full-featured RDBMS, use an actual RDBMS not a minimal-blob-of-code-more-or-less-capable-of-pretending-to-be-a-RDBMS. There's a reason why it's just a few MB of source code rather than a several-hundred-mega install complete with associated daemons and vacuuming cron jobs and whatnot. Anyone told you yet that it stores all data as strings?
That said, I agree that the half-baked ALTER TABLE support is annoying, it would be nice if at least they'd implement their rename-add-copy "recipe" inside SQLite itself so that we don't have to cope with it.
Btw, it seems like in newer versions you now can rename columns.
Finally! Now if we could also change column types that would be awesome.