ah, I see the forum collapses very long threads, and scrolling back it is easy to miss part of my report, responding to Brourd’s comment. If you expand the view more or whatever it is, then at least those screenshots are all preserved. So it looks like I do not have to repost anything.
Therefore, if you would like to see images with more detail, please scroll up, expand the forum posts, and click on any image that you would like higher res on.
My analysis of the roadmap is not on the already documented 30 second network panel, but rather taking a look at the code itself since Brourd kindly pointed out it was particularly long to load the roadmap for my use case.
hmm I tried but failed to edit the above comment, and instead it turned into 2 identical comments, I tried to delete one and that failed and no longer gives me the option.
@whbob thank you very much for these comments. That is very interesting that FF fails to read the metadata - I noticed it explicitly said it was not within the first 1024 chars or something - is it possible you are seeing the declaration, but that it is embedded further down than normal? Or that you have a default rendering that injects a standard when none is declared? Completely guessing here.
Thanks for the note on pressing continue - at the time, nothing happened. I found later this was because even though I had default flash settings ‘ask to enable’, it did not in fact prompt an ask for flash, and therefore failed to load the page even after clicking continue. I don’t know about the metadata at that point. I do know that when I enabled flash to all across the board, then reloaded the page, I was granted access to the puzzle, and don’t recall seeing the error again. I posted screenshots of page loads and puzzle submissions in posts below - you may have to scroll slowly and click view more or whatever it is to expand other forum posts.
Thank you very much for the clarification on avoiding old lab result rendering, this was a really clear phrasing that helped me grasp better how to avoid that particular issue. All my tests and screenshots below are for puzzle submits, logins, and home page loads, and I’m glad to know what to do about avoiding long lab results loads.
[Staff edit: the previous post has been removed per the author’s request]
p.s. I was trying to edit the above comment to clarify that the above posts address your question about expanded and more complete reports for all loading elements, in both FF and Vivaldi Chromium dev tools:
This image is taken out of context from a post above, so the issue may be what I had, which is fast scrolling through forum post misses those frustratingly invisible expand view more links, they are called something else, maybe view replies I don’t know. Anyway, if you scroll back slowly you can see them. Thanks for the tip on sorting by most time, I believe I can do this in Vivaldi since it is based on Chromium. I do not use Chrome and will probably uninstall Vivaldi eventually.
Also, I just tested login and home page again, Win7/Vivaldi, and it took 36 seconds to load home page from newsfeed after a separate 30+ sec login, with the roadmap clocking in at 34 seconds to load, and when sorting all pageload assets by time in the network panel, just confirming that this was the lengthiest option, with no others coming close. Also that ajax-loader.gif file again failed to load.
And just tested submitting tutorial puzzle 8 working up to level 1, first try failed to submit after a couple minutes, reloaded page, sent back to same puzzle undone. Did puzzle, submitted, took 50 seconds to submit. Longest thing to load is post/ at 33 seconds total load time.
Bubbles in background interact with mouse cursor, and also bounce off each other. What calculates and tracks bubble vectors / positions / movement? Is it intensive? Or inconsequential?
Oh, my bad becaus they put here an image of lower resolution and only when I click show the real one, when I drag it to a new tab it looks like it has a lower resolution.
So, I don’t think it has much to do with your machine. I think it is about your internet connection. could you close everything and do and connection speed test? Also, are other sites / games working well for you?
I am going to test RequestPolicy, an addon for firefox, in a bit. maybe you can stop the site from doing noti_count and side project roadmap? then you won’t have some stuff available, but you will still (probably) be able to access puzzle via the menu, and then search for stuff based on the mode you wanted to enter, which is much better than nothing.
i have lost my data 4 times now because I am not used to this keyboard and I keep using a mac shortcut to change windows that blanks out my entry here, so sorry will be short here:
thanks for all suggestions, very good ideas. in brief fyi I’m remembering from top of head since it took me the last 45 minutes to do these tests and I won’t repeat them:
speed test:
66 down
6 up
other games test:
foldit longest puzzle load 8 seconds, otherwise instant nav and load
web games & complex 3D platforms alike have no anomalous behavior
new user in eterna without much data test:
1-2 sec login
2-4 sec homepage load
1.5 sec roadmap load specifically
ajax loader gif not loaded
load labs page 8 seconds
load puzzles page 2-4 seconds
submit puzzle average consistently 9 seconds each for 4 tutorial puzzles
Conclusion is new user much faster login and puzz submit and nav but still all a bit too slow for consistent user engagement
Thank you for the suggestion. This URL only gave text and code, not a rendered page. It may be that this roadmap page is a file without its own internal html, and is only to be loaded by another file that wraps it in its front end UI. On the other hand I am not sure I even understood what I just said, so who knows. Anyone feel free to correct my guesses.
This also held true for trying the eternagame.org domain, and without the user uid.
As for how long it took, the text was an instant load, hooray!
It just prints the data that it gives. My question is, how long does it take there? If you wanna be specific open the network menu and hit F5 then to record exactly.
I did not feel it was necessary to do that, as I said it was fairly instant. Opening up dev tools, and looking at Network feedback, it took 1.69 seconds without and 1.79 with the uid. Why do you want to know? I am not sure I understand, so I will appreciate any context or explanation you can provide, as any hunches may help us further along.
In trying the player puzzles in addition to the 10 step tutorial walkthru, I realized something I had forgotten: that previously the puzzle submission took long enough that I did not wait to see it officially submit, and rather just closed the window and went on to the next.
So, I don’t know whether puzz submit was always very slow and I just got around it by doing this, or if it was a bit slow, but not as bad as it is now. Some folks have noticed things being slower perhaps with certain flash changes.
For example, I came across this while looking at player puzzles:
I do recall in the past of course larger puzzles taking longer even for simply coloring nts.
This also made me realize that the tutorial progression funnels you into waiting for submit completion, rather than being able to close out the tab as described above.
Index Tutorial Puzzles?
Therefore, perhaps a workaround could be making the tutorials available in an indexed link list, like how the labs and player puzzles are all accessible via direct link and indexing.
Being funneled into the guided and gated tour is an understandable UX approach, and I think ( correct me if wrong ) that the gating to access to features would still stand, since it is based on completion of certain puzzles? If there is concern about a user skipping ahead and only completing a few of the latest puzzles to get lab access, for example, then you could set a gate of needing to have completed x amount of puzzles, or even the explicit list of tutorial puzzles or sections ( if it is not already this way ).
Workarounds in Summary
My current workarounds to old users having slow load and submit times, and overheating, are as follows:
Overheating :: Do not use mac laptop, even if it is mbp dual processor with 16GB ram.
Slow Load, Submit, Etc. :: Play another game on another device or have something else to do while waiting.
Slow Player Puzzle Submit :: You don’t have to wait for player puzzles to fully submit, you can close the window and you still get credit & points. I tested this thoroughly yesterday. This may not work on labs or puzzlemaker, I haven’t tried, so copy your sequence before closing a still-submitting page there, to be sure you don’t lose your data. This could be applied to tutorial puzzles, but then you are booted from your ‘next’ button carousel/funnel into the proceeding puzzle, and have to nav back probably thru your home page. Still, if that takes less time than waiting for the submit, at least that may be an option. Note that I haven’t tried this on the tutorial puzzles, so I don’t have a confirmation whether your place in the progression gets submitted or lost, if closing out early.
Slow Tutorial Submit :: Create an indexed direct link tutorial puzzle list so that we can close out without waiting for submit feedback.
That makes 0 sense, although I already reached the conclusion that it will make 0 sense when I did my test of it.
The whole page is the result of your project sidemap or whatever it is called. So, obviously, it calls the server with the request for this data. However, instead of taking 30 seconds, it took 2. Maybe the 30 seconds we see in your screenshot is inflated by multiple requests happening simultaniously?
EDIT:
Wait, could you send the entire url of side project roadmap, including the blacked out stuff? It has 2 things in it- uid, the thing that I can see when I go to your profile, and rnd, which is a random thing for tracking requests as far as I know. Maybe because of rnd it is different, or maybe it has a parameter I don’t know about?
Aha! Thank you so much for the clarity on the purpose for this experiment. Indeed, that is odd, and a great discovery to add to our clue pile.
I would guess the rnd value is a benign factor to force reload of the page with fresh incrementation of roadmap progress ( i.e. not loading precached page with outdated roadmap completion status ), but since I am not a backend programmer, I cannot say for sure, and I seem to recall random numbers also being used for uid login authentication in some systems, but I may be totally wrong here. So I will need to perform an experiment first to confirm, and then may be able to provide this. Alternately, you can experiment on yourself, assuming there is a non negligible difference in load times for you between the two access points, albeit not as severe as my own.
I honestly can’t believe I forgot to mention it, but if you want to be in the chat, you could go to irc.eternagame.org, use the ios app, or use a mirc setup like @hoglahoo.
(p.s. hog said he has problems with irc.eternagame.org, and he is on firefox. so if it takes a suspiciously long amount of time to connect, check if you have an error in the console (f12), because that means it decided to stop trying to connect)
My problem with irc.eternagame.org is most likely related to a firewall according to LFP6 based on some troubleshooting he did for me the other day. Also, I have no trouble with it when I’m on a machine that doesn’t go through the firewall
Another tidbit for you clever people: I reset my profile page to default plain jane and it took 28 seconds to get logged in and fully load the home page in chrome. 28 seconds in firefox as well. changing my profile page back now to all the little frills and doodads since it appears to make a negligible difference in loading times
regarding the _Slow Player Puzzle Submit_, I don’t find it necessary to close the window. Continuing on with the workaround in a new tab seems to be just as fast for me and eliminates any credit or points risk associated with closed the window
While I was investigating the possible cause of Wednesday’s service outage, I stumbled across the fact that the Apache logs were filled with warnings that the connection to the memcached process was failing. To make a long story short, the memcached is a dynamic cache that is intended to provide fast results to common database queries, improving average query times. That process wasn’t running, and apparently hasn’t been running for a long time, because there was nothing about it in the server setup documentation the current devs inherited. So Nando started it up, and I verified that the warning messages went away.
My own impression is that it has actually made a performance improvement. Most noticeably, I haven’t seen any of the intermittent long (~30 second) delays that I have previously observed in the Chrome debugger, where the browser is just waiting for the server to return the first byte of response data.
I am sure that this is not going to fix everyone’s performance issues. But I am curious to see if anyone else thinks there has been some improvement starting about noon today (PDT). It may just be my wishful thinking.
becaus they put here an image of lower resolution and only when I click show the real one, when I drag it to a new tab it looks like it has a lower resolution.
