[DiscordArchive] there is also https://github.com/timelostprototype/wow-client but that code is a mess, i believe its
[DiscordArchive] there is also https://github.com/timelostprototype/wow-client but that code is a mess, i believe its
Original source
there is also https://github.com/timelostprototype/wow-client but that code is a mess, i believe its yoinked from wowser project?
rektbyfaith
07-02-2025, 02:57 PM #1Archived author: Widget • Posted: 2025-07-02T14:57:31.390000+00:00
Original source
there is also https://github.com/timelostprototype/wow-client but that code is a mess, i believe its yoinked from wowser project?
Original source
Yeah, I was thinking about a packet replaying tool. But that approach has its challenges, mainly due to differences in server state between the time of recording and the time of replay. Even if the tool would make some adjustments, it wouldn’t be enough to cover all edge cases or ensure consistent behavior.
So I’d say it’s perfectly reasonable to rely on synthetic load in the tool. After all, even with real players, load patterns can vary significantly. For example 5k players leveling is a completely different load profile compared to 5k players participating in the anqiraj opening event.
I think the most effective approach would be to support multiple behavior/testing scenarios. It would be great if the tool made it easy to define and plug in your own scenarios, so teams could test specific hypotheses about bottlenecks in their setup. But that is something for the far future.
At the end of the day, having synthetic baselines is still valuable. Even if it's not perfect, it provides measurable, repeatable benchmarks and gives a decent starting point for performance tuning.
rektbyfaith
07-02-2025, 05:31 PM #2Archived author: walkline • Posted: 2025-07-02T17:31:23.178000+00:00
Original source
Yeah, I was thinking about a packet replaying tool. But that approach has its challenges, mainly due to differences in server state between the time of recording and the time of replay. Even if the tool would make some adjustments, it wouldn’t be enough to cover all edge cases or ensure consistent behavior.
So I’d say it’s perfectly reasonable to rely on synthetic load in the tool. After all, even with real players, load patterns can vary significantly. For example 5k players leveling is a completely different load profile compared to 5k players participating in the anqiraj opening event.
I think the most effective approach would be to support multiple behavior/testing scenarios. It would be great if the tool made it easy to define and plug in your own scenarios, so teams could test specific hypotheses about bottlenecks in their setup. But that is something for the far future.
At the end of the day, having synthetic baselines is still valuable. Even if it's not perfect, it provides measurable, repeatable benchmarks and gives a decent starting point for performance tuning.
1 Guest(s)
1 Guest(s)