|
git --out-of-retail luigi-thirty
|
# ? Dec 16, 2014 15:30 |
|
|
# ? May 24, 2024 07:35 |
|
vodkat posted:And what else is there? opera 15 and later are lovely forks of chrome they fired the entire opera dev team, so any features you miss from opera 12 are not gonna show up unless google decides to add them
|
# ? Dec 16, 2014 16:01 |
|
c c# s: parallel.foreach is my favorite fuckin thing
|
# ? Dec 16, 2014 16:14 |
|
is it bad that my current scheme for handling database lock errors is basically wrapper(query) try { query.execute } catch { wrapper(query) } n just hoping the database issue resolves itself before the stack overflows
|
# ? Dec 16, 2014 16:16 |
|
Bloody posted:is it bad that my current scheme for handling database lock errors is basically if the program failing doesn't cost a company a large amount of money its fine i'd say
|
# ? Dec 16, 2014 16:17 |
|
rotor posted:if you use git i have no idea why you wouldn't use gitlab or hub or whatever gitlab and github are different things for solving the same problem
|
# ? Dec 16, 2014 16:18 |
|
Notorious b.s.d. posted:opera 15 and later are lovely forks of chrome yeah firefox is pretty okay imo, same for IE it has been doing better recently
|
# ? Dec 16, 2014 16:19 |
|
Bloody posted:c c# s: parallel.foreach is my favorite fuckin thing C# is cool and good.
|
# ? Dec 16, 2014 16:21 |
|
Bloody posted:is it bad that my current scheme for handling database lock errors is basically I'm guessing a backoff mechanism would usually be nice for this stuff, and/or giving up after a while if possible. The actual issue there is probably that there's no clearly defined semantics of what happens when the DB is truly unavailable and that requests would fail. So what's the path to recovery or telling the user something failed? Is there any one? In the real world, "we just hope this doesn't fail" is poor planning. What if the front-end server dies while the DB is locked up? Is the query built in a way such that it can be retried even if it succeeded but the user didn't know about it because of ~issues~? Those are fairly hard questions, but if you can manage to answer them, the rest of it kind of falls in place on its own. They're fixing the hardest problem for your software: what happens when nothing works, and what can a user do? Then you can, in the worst case, reuse that solution when something only works partially, or doesn't work very well. That gives you a lower bound on system awfulness from the user's perspective. If you've taken or planned no measure for that, then yeah, you probably have no bound on how awful the experience will be for a user when things go bad, and most of the stuff around other failure cases will just delay the inevitable.
|
# ? Dec 16, 2014 16:23 |
|
Careful Drums posted:gitlab and github are different things for solving the same problem from a user perspective I don't see a significant difference between the two. one uses branches one uses clones repos, that's about the extent of the difference as far as I could tell.
|
# ? Dec 16, 2014 17:26 |
|
triple sulk posted:C# is cool and good. it's true
|
# ? Dec 16, 2014 17:26 |
|
Bloody posted:is it bad that my current scheme for handling database lock errors is basically how do you have so many locking problems are you on mysql or something
|
# ? Dec 16, 2014 18:21 |
|
iirc hes using SQLite but somehow loving up locks which seems like the weirdest thing.
|
# ? Dec 16, 2014 19:04 |
|
sqlite and those old issues are gone now this is somewhat unrelated
|
# ? Dec 16, 2014 19:10 |
|
im not even getting "a lot" of locking issues, just a few ever. im doing a bunch of poo poo in parallel and the db isn't the bottleneck but i can occasionally time out while inserting a bunch of crap so this way it just retries until it succeeds instead of failing miserably actual reliability isnt all that important
|
# ? Dec 16, 2014 19:11 |
|
rotor posted:from a user perspective I don't see a significant difference between the two. one uses branches one uses clones repos, that's about the extent of the difference as far as I could tell. the goal of gitlab is to be a feature-equal almost-clone of github, and then they add some additional abilities, like you can have pre-push hooks on your repos
|
# ? Dec 16, 2014 19:12 |
|
oh man I just remembered how much I hate having to configure hooks as a loving web service
|
# ? Dec 16, 2014 19:16 |
|
wait how did i get here
|
# ? Dec 16, 2014 19:19 |
|
Shaggar posted:like there are legitimate reasons for hazing your junior staff like that. Working support while doing development on the software gets you experience and understanding with both sides of the product which is very valuable. who knows, they could just be smart people and you come out at the end of the year a better developer who understands your customer's needs. had my phone interview, this was what they said. also i was able to answer all their C#, SQL Server, data structures questions, plus a Java question that I guessed at because I am a java nooblord also they want to see me next week for an in-person interview, gently caress retail forever
|
# ? Dec 16, 2014 19:32 |
|
no matter how bad things get, at least i will always be better than the people who use opera to browse the web
|
# ? Dec 16, 2014 20:31 |
|
opera was cool back when it used to be the only browser available that could render mobile web pages
|
# ? Dec 16, 2014 20:35 |
|
BONGHITZ posted:no matter how bad things get, at least i will always be better than the people who use opera to browse the web
|
# ? Dec 16, 2014 20:42 |
|
Luigi Thirty posted:had my phone interview, this was what they said. also i was able to answer all their C#, SQL Server, data structures questions, plus a Java question that I guessed at because I am a java nooblord
|
# ? Dec 16, 2014 20:46 |
|
Valeyard posted:opera was cool back when it used to be the only browser available that could render mobile web pages i proudly posted on the forums with Opera Mini on a Blackberry Curve 8320 all through high school
|
# ? Dec 16, 2014 20:47 |
|
Luigi Thirty posted:had my phone interview, this was what they said. also i was able to answer all their C#, SQL Server, data structures questions, plus a Java question that I guessed at because I am a java nooblord u can do it!!
|
# ? Dec 16, 2014 20:58 |
|
Luigi Thirty posted:had my phone interview, this was what they said. also i was able to answer all their C#, SQL Server, data structures questions, plus a Java question that I guessed at because I am a java nooblord nice, good luck
|
# ? Dec 16, 2014 21:00 |
|
I believe in you luigi30
|
# ? Dec 16, 2014 21:58 |
|
Bloody posted:im not even getting "a lot" of locking issues, just a few ever. im doing a bunch of poo poo in parallel and the db isn't the bottleneck but i can occasionally time out while inserting a bunch of crap so this way it just retries until it succeeds instead of failing miserably seriously, own up, what in gently caress are you doing with mutli writers to sqlite i mean, heck, in python, you can't even use it with threads
|
# ? Dec 17, 2014 00:25 |
|
Luigi Thirty posted:had my phone interview, this was what they said. also i was able to answer all their C#, SQL Server, data structures questions, plus a Java question that I guessed at because I am a java nooblord get a friend to run a practice interview
|
# ? Dec 17, 2014 00:25 |
|
tef posted:get a friend to run a practice interview ok, first question: where is the $30 you owe me?
|
# ? Dec 17, 2014 00:28 |
|
tef posted:seriously, own up, what in gently caress are you doing with mutli writers to sqlite Basking in the glory of this thread
|
# ? Dec 17, 2014 00:44 |
|
i'm maintaining some code that has a triple race condition. but it's not really a race condition, because it's not one process failing because it ran before something else could complete. it's failing because a single code path calls the same asynchronous process 3 times, and they frequently execute simultaneously. those asynchronous processes happen to be http posts, so they're extra asynchronous. anyhow they all try to post the same thing with the same data and then they all partially succeed and then restart themselves and wow is it a mess. i added a mutex and that's making things better but that only solves the problem if the processes execute simultaneously. they also sometimes execute subsequently and the mutex does nothing for that. i could add a flag that says 'hey, you already did this, dont do it again' but i'm 90% sure that would gently caress with other logic in the system. the real answer is to unfuck the codepaths, because there's no legitimate reason for this to be happening 3 times, but unfucking the codepath would definitely break a lot of other things
|
# ? Dec 17, 2014 02:14 |
|
if its not supposed to execute 3 times why wouldn't eliminating 2 requests fix it?
|
# ? Dec 17, 2014 02:18 |
|
Shaggar posted:if its not supposed to execute 3 times why wouldn't eliminating 2 requests fix it? because the code is really really bad and other things are very likely to depend on the side effects of this process i also cant assume things like 'hey, this object says it's already been posted to the remote services, so there's no need to submit it again' because i cant guarantee that there isn't something that relies on clearing attributes on the object and resubmitting it DONT THREAD ON ME fucked around with this message at 02:25 on Dec 17, 2014 |
# ? Dec 17, 2014 02:22 |
|
MALE SHOEGAZE posted:because the code is really really bad and other things are very likely to depend on the side effects of this process i'm sure that'd show up in the tests though
|
# ? Dec 17, 2014 02:32 |
|
Stringent posted:i'm sure that'd show up in the tests though the ppl who programmed this before me set up the workers so that they dont run in staging. ever
|
# ? Dec 17, 2014 02:33 |
|
MALE SHOEGAZE posted:the ppl who programmed this before me set up the workers so that they dont run in staging. ever can you give your postings some kind of unique identifier such that the remote service can ignore the postings it has already received
|
# ? Dec 17, 2014 03:01 |
|
bobbilljim posted:can you give your postings some kind of unique identifier such that the remote service can ignore the postings it has already received no but i can get the data and then not post if it's already present
|
# ? Dec 17, 2014 03:02 |
|
rotor posted:I don't understand why people use github in a world where gitlab exists gitlab perf can be pretty variable, but it got much better in recent versions.
|
# ? Dec 17, 2014 03:05 |
|
|
# ? May 24, 2024 07:35 |
|
eschaton posted:gitlab perf can be pretty variable, but it got much better in recent versions. it's almost as if there's no silver bullet because if there was everybody would be using it already
|
# ? Dec 17, 2014 03:10 |