EVE Search - Announcement:Skill Queue Issues

FlySilver wrote:

Rexxar Santaro wrote:

Lquid Drisseg wrote:

I am a software developer. This is not a hard problem to solve. They already have it solved. Skill training is already a thing. It would not be hard to add in a base level of training that occurs regardless of your queued skills and attribute levels.

Please don't claim this is a hard problem, especially when the vast majority of the actual hard work has already been implemented.

IGÇÖm pointing that the server multithreading programing is very complicated and anything missed can turn into huge problems. ItGÇÖs far away not like to client desktop or mobile apps, but nm.

The problem is most of those threads arenGÇÖt obviously predictable, based on some linear function, which you can apply additive to all data. ItGÇÖs mostly random, which should take into consideration what everyone did the last 10 hours, including subscription gain/lost, plexes, skill injection, skill extraction, attribute changes. Maybe somebody get podded and was for 1 hour at station without any attribute implants active you know. It requires to analyze a lot of log records and recreate the story based on them.

I offered to grant to everyone a micro SkillInjector, as average SP for all accounts. Almost all will be happy and CCP will not waste time on this.

I'm pointing that you obviously have no idea about multithreading programing. If you think it is very complicated you most likely have no experience in programming at all. I would even go as far as to say that all of the threads are predictable. The Reason it is that way, is "obviously", you code that Threads yourself. If you have to rely on other Threads you dont have written yourself you do not in any way expect those Threads to behave at all and work with them in a completely different way that has nothing to do with "multithreaded programing" in its basic meaning.


Author	Thread Statistics \| Show CCP posts - 1 post(s)
Rexxar Santaro Republic Military School Minmatar Republic 26	Posted - 2017.06.09 13:21:44 - [1] - Quote It happened, looks like, at early morning of 2017-06-09. Maybe 10 hours passed since. BTW, yesterday I encountered high latency and huge lags on some UI between 21:00 -22:00 ST. Just grant to everyone a micro SkillInjector :)...
Rexxar Santaro Republic Military School Minmatar Republic 26	Posted - 2017.06.09 14:07:07 - [2] - Quote Lquid Drisseg wrote: This seems like a really really simple problem to fix going forward. Why cant we just make this simple for everyone? If you have skills in a queue, they train at the attribute level rate. If you "pause" your skillqueue or otherwise are not training anything and still subbed, just give accounts a calculated fair SP rate passively. Just make SP accumulation a passive thing that you apply in chunks as you want to train skills. ItGÇÖs very hard, a nightmare, for a developer to program and synchronize both CPU-server with a DB-server, considering every skill queue is a server real time thread (with own list of different skills). Also, while everyone can interact with that thread through: subscriptions, plexes, attributes, implants, skill extractors, skill injectors. For sure there are running over 100,000 live threads! Dynamic multithreading server programing is among hardest types. One bit of leaked memory (TRUE or FALSE) can turn into hundreds of Gigabytes of leaked memory in one hour. As a result GÇô server shutdown or Try Catch exception, which will stop the problematic code. To track and count the SP for everyone is possible after processing a gazzilion of data.
Rexxar Santaro Republic Military School Minmatar Republic 26	Posted - 2017.06.09 14:40:38 - [3] - Quote Firicel wrote: I want my SP back! Dude, looks like you aren't happy about what happened!
Rexxar Santaro Republic Military School Minmatar Republic 26	Posted - 2017.06.09 15:00:49 - [4] - Quote Lquid Drisseg wrote: I am a software developer. This is not a hard problem to solve. They already have it solved. Skill training is already a thing. It would not be hard to add in a base level of training that occurs regardless of your queued skills and attribute levels. Please don't claim this is a hard problem, especially when the vast majority of the actual hard work has already been implemented. IGÇÖm pointing that the server multithreading programing is very complicated and anything missed can turn into huge problems. ItGÇÖs far away not like to client desktop or mobile apps, but nm. The problem is most of those threads arenGÇÖt obviously predictable, based on some linear function, which you can apply additive to all data. ItGÇÖs mostly random, which should take into consideration what everyone did the last 10 hours, including subscription gain/lost, plexes, skill injection, skill extraction, attribute changes. Maybe somebody get podded and was for 1 hour at station without any attribute implants active you know. It requires to analyze a lot of log records and recreate the story based on them. I offered to grant to everyone a micro SkillInjector, as average SP for all accounts. Almost all will be happy and CCP will not waste time on this.
Rexxar Santaro Republic Military School Minmatar Republic 26	Posted - 2017.06.09 15:26:50 - [5] - Quote Alura Eidolon wrote: I'm not sure giving out Skill injectors is the way to go mainly because of the variance. Someone with less than 5mil SP gets alot more of a boost than someone with over 80mil SP. Plus then you are getting a free skill injector that you can sell on the market making a profit for, potentially, a lot more SP than you lost. Well, this special micro Skill Injector must have the average amount of SP which can be trained during the 10 hours (8hours IDK). The injector should have blocked trading attributes: unable to trade on market or through contracting and must be connected to account like that starting one. Like some active goods have. The amount of gained SP must be static GÇô independent from currently capsuleer SP amount.
Rexxar Santaro Republic Military School Minmatar Republic 26	Posted - 2017.06.09 20:25:54 - [6] - Quote FlySilver wrote: Rexxar Santaro wrote: Lquid Drisseg wrote: I am a software developer. This is not a hard problem to solve. They already have it solved. Skill training is already a thing. It would not be hard to add in a base level of training that occurs regardless of your queued skills and attribute levels. Please don't claim this is a hard problem, especially when the vast majority of the actual hard work has already been implemented. IGÇÖm pointing that the server multithreading programing is very complicated and anything missed can turn into huge problems. ItGÇÖs far away not like to client desktop or mobile apps, but nm. The problem is most of those threads arenGÇÖt obviously predictable, based on some linear function, which you can apply additive to all data. ItGÇÖs mostly random, which should take into consideration what everyone did the last 10 hours, including subscription gain/lost, plexes, skill injection, skill extraction, attribute changes. Maybe somebody get podded and was for 1 hour at station without any attribute implants active you know. It requires to analyze a lot of log records and recreate the story based on them. I offered to grant to everyone a micro SkillInjector, as average SP for all accounts. Almost all will be happy and CCP will not waste time on this. I'm pointing that you obviously have no idea about multithreading programing. If you think it is very complicated you most likely have no experience in programming at all. I would even go as far as to say that all of the threads are predictable. The Reason it is that way, is "obviously", you code that Threads yourself. If you have to rely on other Threads you dont have written yourself you do not in any way expect those Threads to behave at all and work with them in a completely different way that has nothing to do with "multithreaded programing" in its basic meaning. Heh, this thread isnGÇÖt about examining somebodyGÇÖs multithreading programing knowledge. Somebody is a good shader programmer, other one is a good multithreading programming and other one is good in developing huge DB-systems. Shortly, letGÇÖs put apart simple client multithreading programs (when you just open a file or compressing it in a thread and then it is closed), client dynamic threads and server multithreading systems (where every running thread works into an infinite loop and should interact with many different systems: web-server, client exe, mobile app, DB-Server and etc; the problem of secured thread programing and memory leaks is well known). Compared to OOP, Delegate programing or Interfacing concepts the threading one is more complicated because of this Debugging process at least. In big projects itGÇÖs OMG when somethingGÇÖs going wrong, because a well working thread can suddenly work wrong just because the imported or exported data form/to DB, Web, EXE can have an unexpected value/format/structure which will create exceptions or bad logical decision that are hard to detect. The number of bug fixes and patches, in every serious project available on market, are just an example that this isnGÇÖt simple. P.S. Sorry about this wall of text...