The new browser is not ready to take over

  • 5
  • Problem
  • Updated 3 years ago
  • In Progress
  • (Edited)

We need the classic browser back around until the new browser is actually capable of allowing us to perform the same tasks as before, when we design and do analysis.


The classic browser is stress tested during many years. Nando and Jnicol have fixed its biggest problems so it is now supporting our work in an easy manner. This is not yet the case for the new browser.


While the new browser has cool new options, many of the most basic functions are still missing.


We need to be able to see the design structure - else we can’t make good use of the results from the earlier round to improve designing from this round.


We need to be able to view design description and copying out sequence to modify designs and be able to experiment on them. All these are things that are crucial for designing and analysis. This not being possible, gets in the way of us making good use of the lab results we have gotten back.


This is a problem!


Please come with your input. What options do you miss? What things would you like to see? Lets help Omei work out the wrinkles.
Photo of Eli Fisker

Eli Fisker

  • 2253 Posts
  • 506 Reply Likes

Posted 4 years ago

  • 5
Photo of Eli Fisker

Eli Fisker

  • 2253 Posts
  • 506 Reply Likes
Wish number 1  - seeing the design structure


http://www.eternagame.org/web/labs/data-browser/?puzzle_id=6296750

For many of the latest archived labs, it is not possible to see the 2D designs structure. We depend on them for analysis. Without the design structure, we can’t learn from past results to improve on current round.


This is already a problem as I now see even fewer of the new lab designers use past results for designing. We get a lot of from scratch design - which is good - but without being able to see and learn from the earlier results, designing is blind and we are not benefiting as we should from our precious lab results.


2D design structure as seen in the classic browser

http://www.eternagame.org/game/browse/6296750/?filter1_arg2=6320914&filter1=Id&filter1_arg1=...
(Edited)
Photo of mat747

mat747

  • 130 Posts
  • 38 Reply Likes
I agree with Eli the 2D design structure is still needed for analysis.
Photo of Omei Turnbull

Omei Turnbull, Player Developer

  • 989 Posts
  • 310 Reply Likes
OK, I've re-implemented the 2D structure view so that it makes use of the existing flash app, and thus works for all designs.  It may not be perfect, but think this removes the design structure as a show stopper.  What do you all think?
Photo of LFP6

LFP6, Player Developer

  • 618 Posts
  • 109 Reply Likes
I think it would help to have a more lightweight interface eventually. Having the entire UI in the window seems overkill. Perhaps if there was a way to have it just show basic information (ie the structure/data) by default, then switch to an "experimentation" view (ie opening withing a full game window or something)?
Photo of Eli Fisker

Eli Fisker

  • 2253 Posts
  • 506 Reply Likes
The structure boxes do not adjust to the space available, making it hard to see the second one.

Photo of Eli Fisker

Eli Fisker

  • 2253 Posts
  • 506 Reply Likes
Each time I am shifting from one design to a new design there is a waiting time on 15-20 seconds. The structure image is loading painfully slow.
Photo of Omei Turnbull

Omei Turnbull, Player Developer

  • 989 Posts
  • 310 Reply Likes
Is it always 15-20 seconds?  Downloading  the flash app can take that long, but the Eterna server is now telling the browser that it is OK to cache the app.  (That is, the browser just does a quick check to make sure the app hasn't been updated and then loads the cached copy if it hasn't been.)  I see typical wait times of 4-5 seconds for a new 2D instance.

If you are always having a 15-20 second wait, I would want to know if your browser is for some reason not caching the app, or whether something else is wrong.  I would suggest starting by loading the app in Chrome and bringing up it's debugger, because I think it is the easiest to use.  Do you know how to do this?
Photo of Omei Turnbull

Omei Turnbull, Player Developer

  • 989 Posts
  • 310 Reply Likes
@Eli: Re the 2D structure boxes, I've coded a change on http://www.eternadev.org/lab/data-browser.html to maintain a 3:2 aspect ratio. What do you think about that version?
Photo of Eli Fisker

Eli Fisker

  • 2253 Posts
  • 506 Reply Likes
Hi Omei!

Big thx. This is far better.

It still is a problem when I pull 3 windows.
Photo of Eli Fisker

Eli Fisker

  • 2253 Posts
  • 506 Reply Likes
Yes, it is not just delay the first time I load the structure drawing. I'm trying to change my browser settings so they allow better for saving of data.

I usually have no problems in the classic browser, there load time is less than a sec.

I don't know how to load the debugger in chrome. I will need some guidance.
Photo of Omei Turnbull

Omei Turnbull, Player Developer

  • 989 Posts
  • 310 Reply Likes
Re "It still is a problem when I pull 3 windows.":  You control how many of the structure boxes you see by adjusting the width of the right panel.  Of course, the more you put there, the smaller they are.

BTW: Something you'll come across sooner or later: if you drag the left border of the right panel quickly while the new border is over the 2D boxes , it can appear to get stuck.   If this happens, you just need to move the border more slowly.  Undoubtedly, there is something I can do to improve this, but I'm not thinking of it as a priority right now.
Photo of Omei Turnbull

Omei Turnbull, Player Developer

  • 989 Posts
  • 310 Reply Likes
@Eli, re the debugger, I've created the document https://docs.google.com/document/d/1uLyORTRpZsSbmJrtbEiI3gOwmuVRJNXjG7My_5jgF7Q/edit#heading=h.pgzu7... and given you edit permissions.  We can continue the discussion there.

@others: The document should be open to viewing and comments, in case the topic is of interest to you.
Photo of Eli Fisker

Eli Fisker

  • 2253 Posts
  • 506 Reply Likes
Stability - columns and headlines


The new browser is not yet stable when it comes to displaying the results.



A known and recurring problem is that column headers slide in relation to the columns. This makes reading out results far more tedious than it has to be.
Photo of Omei Turnbull

Omei Turnbull, Player Developer

  • 989 Posts
  • 310 Reply Likes
Eli, can you tell me how to reproduce this particular bug?
Photo of Eli Fisker

Eli Fisker

  • 2253 Posts
  • 506 Reply Likes
It happens regularly, typically when I have been setting something in the left panel. I will try to look out for when exactly it happens.
Photo of Eli Fisker

Eli Fisker

  • 2253 Posts
  • 506 Reply Likes
I just replicated the sliding error. All I did was open this lab:


Slide replicatiompng
http://www.eternagame.org/web/labs/data-browser/?puzzle_id=6296745

I'm using firefox.
Photo of Omei Turnbull

Omei Turnbull, Player Developer

  • 989 Posts
  • 310 Reply Likes
Thanks Eli, I can reproduce it now.  It looks like it is specific to Firefox and having the Sequence column displayed.  (Headers align correctly if I hide the Sequence field.) If you come across any other example that doesn't have both these elements present, please post again.

I'm not sure how long it will before I get it fixed.  :-(
Photo of Eli Fisker

Eli Fisker

  • 2253 Posts
  • 506 Reply Likes
Copy out sequences



When I design for lab, I regular modify or do experiments to a design from the earlier round. To do this I need to be able to copy the sequence out from the archived data. The new browser have no sequence copy function.


Photo of LFP6

LFP6, Player Developer

  • 618 Posts
  • 109 Reply Likes
This does still exist, it is only available via the in-game browser (accessed via the hamburger menu while in the puzzle). Ideally this should be available via the web browser, I agree.
Photo of Eli Fisker

Eli Fisker

  • 2253 Posts
  • 506 Reply Likes
My problem is that the Lab archive that I go to as usual, via the hamburger menu, for a lot of the past labs now sends me directly into the new browser and not the classic browser where the option is available.
(Edited)
Photo of LFP6

LFP6, Player Developer

  • 618 Posts
  • 109 Reply Likes
For all the ones I tried, this worked:




Photo of Eli Fisker

Eli Fisker

  • 2253 Posts
  • 506 Reply Likes
Thx LFP for finding a backdoor that doesn't take url editing.

Usually we can get to the designs directly by taking the Sequence button, without having do do things backwards.
Photo of LFP6

LFP6, Player Developer

  • 618 Posts
  • 109 Reply Likes
Actually, when you hit sequences before, it still wasn't the same one! If you remember from quite a while back, there only used to be the Flash browser, then they introduced one in HTML5. The Flash one stuck around within the game, and the HTML5 one was accessible via the sequences button. This one has been replaced, but not the one in-game.
Photo of Rocketdog42

Rocketdog42

  • 33 Posts
  • 2 Reply Likes
I've been doing the same thing as Champion.
Photo of Eli Fisker

Eli Fisker

  • 2253 Posts
  • 506 Reply Likes
Copy out link



When I refer to a design I modify, I typically leave a link back to the original design in my design description.


Leaving a link to the design I get inspired from is both a courtesy to the original designer, but also a help for analysis when we get the data back. Then I can easier see if what I intended really did work compared to what I started from.


Idea: If copying out sequence and referencing URL, and original score could be done in one go - in other words supported by the interface, that would be even better. We still need to be able to get either of these individually.

(Edited)
Photo of LFP6

LFP6, Player Developer

  • 618 Posts
  • 109 Reply Likes
From what I understand, design ancestors is now tracked for more recent puzzles (ie when you copy a design, it should record what you copied it from via the in-game browser) but this information is not displayed in the interface itself. I could probably script up a viewer that displays the entire ancestry as a tree kind of thing if I get the chance. Of course this would ideally be integrated more into the interface, both for displaying the history and navigating to the data for any of the related designs.
Photo of LFP6

LFP6, Player Developer

  • 618 Posts
  • 109 Reply Likes
Actually I take that back. It doesn't appear that the data is available via the API at this point. Hopefully it will be available soon.
Photo of Eli Fisker

Eli Fisker

  • 2253 Posts
  • 506 Reply Likes
Hi LFP!

Just wanted to say thanks for giving it a go.
Photo of LFP6

LFP6, Player Developer

  • 618 Posts
  • 109 Reply Likes
At least the information is being tracked (at least as long as it's being stored, I do know that t's being sent), so it is there to be used.
Photo of LFP6

LFP6, Player Developer

  • 618 Posts
  • 109 Reply Likes
And again, the link copying feature is still there in the in-game browser.
Photo of whbob

whbob

  • 198 Posts
  • 66 Reply Likes
Omei and the developers in general are doing a great job. While they continue the Reorg. of the browser, I've used a work around to access past labs. 

LABS, Project ID’s, puzzle ID’s Design(solution) ID’s.

Example LAB: Next Generation Riboswitches On Chip, (R1). 

Project ID# 5851792.

Sublabs:

Exclusion  NG 1 Puzzle ID# 5851774.

Exclusion  NG 2 Puzzle ID# 5851780.

Exclusion  NG 3 Puzzle ID# 5851784.

Same State NG 1 Puzzle ID# 5851785.

Same State NG 2 Puzzle ID# 5851787.

Same State BG 3 Puzzle ID# 5851791.


The Project ID# is the alternate description of the Lab Name.  

This lab contains 6 sub-labs.  

The puzzle ID# is the alternate description of the sub-lab name.   

The sub-lab has a list of Design iD’s for that sub-lab.  

The Design id’s are the individual lab shape designs for the sub-lab.


Summary:

Project ID#, Puzzle ID#, Design ID# is the hierarchy.

Original (old) URL format:

www.eternagame.org/game/browse/Project ID#/. 

Goes to the Lab description page which should include the subl-lab icons.


www.eternagame.org/game/browse/puzzle ID#/. 

Goes to a list of the design ID#’s in the puzzle ID sub-lab. 

 

The “/“ symbol must be at the end of the URL for a successful search.


It would be helpful if there was a spreadsheet to reference past labs to ID numbers :)

Photo of Eli Fisker

Eli Fisker

  • 2249 Posts
  • 503 Reply Likes
Neat Whbob!

I did notice that you were one of the few who did manage to get to and use the past lab results. :)
Photo of Omei Turnbull

Omei Turnbull, Player Developer

  • 989 Posts
  • 310 Reply Likes
For the record, I completely agree that the new browser is not ready to take over from the legacy (Flash) one.  I certainly still use both, even for labs that have data in the new browser's database.

I am currently working with Nando to get the 2D structure into the browser by embedding a reduced-functionality version of the flash applet there.  I have two different UI possibilities in mind and could use some player feedback.  But I'll post that in the https://getsatisfaction.com/eternagame/topics/new-lab-data-browser-for-eterna-medicine topic.
Photo of Eli Fisker

Eli Fisker

  • 2253 Posts
  • 506 Reply Likes
I have gotten tired of having to ticking off lots of boxes each time I want to see data for a different lab - something which happens a lot - so I'm thinking that it would be really helpful if the browser could just remember my global preferences. Just as the classic browser remembers which columns I wish to have shown globally. It is usually the same data columns I wish to see no matter the lab.

Ok, really this solution is probably less good than just having all the favored columns remembered by default.


 
Photo of Omei Turnbull

Omei Turnbull, Player Developer

  • 989 Posts
  • 310 Reply Likes
Remembering past preferences has been an important design consideration from the beginning (ask Siqi :-) ), and I am puzzled why it isn't working for you, Eli.

Note that your preferences are stored in what HTML5 calls "local storage".  Is it possible that you routinely work in "incognito" mode?  If so, that would account for your preferences not being saved. The HTML spec requires that the browser clears out local storage data when the current session terminates. 

If that isn't your issue, let me know and we'll figure out what is going wrong.
Photo of Eli Fisker

Eli Fisker

  • 2253 Posts
  • 506 Reply Likes

Less data viewable in the web browser compared to the flash browser


While the classic browser has way too much space in its data columns, I can still see far more both of the data columns and the sequences per lab. Which I like.


Flash, 40 sequences, sequence list fills 2/3

Amount of data flashpng


In the web browser, the sequences fills more and I can see fewer off them. This leaves me with less data columns on the side.


Web, 30 or really 29 sequences, sequence list fills 3/4

Web overviewpng


What is the greatest promise of the web browser is if one could view the sequence list alongside a design structure drawing and still have room for one or two data columns. The sequences filling too much leaves very little room for this. All extra space would be helpful for making the structure drawing bigger. 


Full viewpng


Some of the versions of the Data Browser have held the 40 sequence view. And I really liked it.

Photo of Omei Turnbull

Omei Turnbull, Player Developer

  • 989 Posts
  • 310 Reply Likes
The main difference between your first two screenshots is the zoom level, and you have full control over that with the Control +/- keys.  (At least that is what they are for Chrome; I think it is pretty standard across browsers.)

Am I missing something?
Photo of Eli Fisker

Eli Fisker

  • 2253 Posts
  • 506 Reply Likes
I have zoom set to 100% in the browser for both images.

This happens in both firefox and chrome, but I can only shoot this image including zoom level within firefox.

Firefox with zoompng
Photo of Omei Turnbull

Omei Turnbull, Player Developer

  • 989 Posts
  • 310 Reply Likes
My point was not that everything is the same in both browsers at the same zoom level, but that the player has control of the zoom level.  There's an obvious trade-off between how much you can get on the screen and how well you can see it because of reduced size.  That choice is a personal one, based on screen size and resolution, one's vision, what other windows one also wants to see, etc.

Making the best use possible of whatever space is available is definitely a design goal.  But trying to tweak the best compromise for a default zoom level doesn't seem like a good use of time unless there is a consensus among players that the current one is clearly sub-optimal.
Photo of Eli Fisker

Eli Fisker

  • 2253 Posts
  • 506 Reply Likes
I have grown particular fond of the kcal column lately. I know the classic browser shows the Vienna value and the game shows the NUPACK value. 

However I can't find the nupack one in the spreadsheets and as such it doesn't get into the web browser.

I would like to get kcal into the web browser. I have found it helps me for analysis.
Photo of LFP6

LFP6, Player Developer

  • 618 Posts
  • 109 Reply Likes
In case it's of interest, based on some exparamentation I did, it's actually the energy from the state and energy model you submitted it with.
Photo of Eli Fisker

Eli Fisker

  • 2253 Posts
  • 506 Reply Likes
LFP, thx. This is very relevant.

What I understand you are saying is that the column kcal can then be from different models then.

And the kcal on the design shown by default is NUPACKS - for most of the latest labs, we have NUPACK as default once we get the data back and also when designing.

So having kcal from each engine split up in columns would be most helpful. Or having it consistently be from NUPACK. It will help for comparison if we can compare the same energy model and not different ones in the same column.
Photo of Omei Turnbull

Omei Turnbull, Player Developer

  • 989 Posts
  • 310 Reply Likes
It's my current plan to not use the energy values stored in the Eterna database for the new browser.  Instead, the values would be generated by the same process that generates the dot-bracket structure strings, so that they are consistent.
Photo of Eli Fisker

Eli Fisker

  • 2253 Posts
  • 506 Reply Likes
Crucial information missing

I have found that the small boxes with info about which switch elements needs to be present in what state are not there. It holds great value to know, both to see if the puzzle fulfills demands - also of the simulation. At the same time this teaches us about the quirks of NUPACK. When we can see when lab allow things that simulation disagrees with. I'm in chrome.

Small info boxespng

The same puzzle in the classic browser

classic browserpng
Photo of Eli Fisker

Eli Fisker

  • 2253 Posts
  • 506 Reply Likes
Sorting function missing



One of the tools I value the most in the classic browser is the sorting function. It's invaluable when it comes to distill what is working.


What one can do is click on any design of interest and get the data sorted by their similarity to this design.


It is by far the tool that tells me most, when it comes to comparing similar designs and see which small base changes are the most important and why. However it is not yet an option in the new browser.


I have earlier given an intro to the sorting function. The same document contains ideas for how to use it.
Photo of whbob

whbob

  • 198 Posts
  • 66 Reply Likes
Can't get to the web site today.  Just a blue screen.  I can see the new lab page on the dev server.  Looks like 4 or 5 oligo's.  Delay looks to be about 6 million miles at light speed :)  5/3/16  
Photo of whbob

whbob

  • 198 Posts
  • 66 Reply Likes
Just got back on.  Safari was my problem browser.  Switched to chrome yesterday and it worked great.  Today, safari is back up and running again.  Thanks to all who worked on this :)
Photo of whbob

whbob

  • 198 Posts
  • 66 Reply Likes
I was just looking at a Gene synthesizer puzzle.  NT numbers didn't show on the RNA strand when it first opened.  Neither natural or target mode.
I opened the settings and Show Numbers was checked.  I unchecked the box, went back to the puzzle, than back to the settings.  Rechecked the NT numbers selection.  The numbers then appeared in the puzzle OK.  
New players have a hard time when the numbers are not shown.
Photo of whbob

whbob

  • 198 Posts
  • 66 Reply Likes
Here is a png of my browser(Chrome)with an error report.  The 2D section of the browser never loads the Flash structure.  This began after Flash had a new update recently.  I'm guessing that the 2D area is a frame that calls the Flash software.  Not sure where the call to flash is in the code, but this error may be related to that call not being processed properly.  Hope this helps :)

Photo of Omei Turnbull

Omei Turnbull, Player Developer

  • 988 Posts
  • 310 Reply Likes
Thank you, whbob, for the detailed screenshot.

The problem doesn't seem to be your flash version, but the fact that puzzleid is undefined in the Puzzlequery string. Your screenshot doesn't show the main table, but offhand the only reason I can think of for the puzzle id being undefined is that you didn't start by choosing one or more puzzles.  Is this a possibility?  If not, maybe you can give me the full sequence of what you do to create this.  Or is this not a easily reproducible error?
Photo of whbob

whbob

  • 198 Posts
  • 66 Reply Likes
Ask not if "The new browser is ready to take over?", ask " Am I ready for the new browser?"  
I saw the same thing last night.  "Puzzleid not defined", using dev tools in chrome.  I woke at the break of dawn thinking " the filters".  
I filtered out all of the "unnecessary data on the puzzle lines, thinking it was only a visual for me and not a filter of the data that the software would use.
I listed eterna scores greater than 89 with a sublab puzzle name showing the design, score and sequence, not thinking something like the puzzle id was necessary. 
Went back and checked puzzle id and now all is well in "my" browser .

Sorry for the distraction :(  
Photo of Omei Turnbull

Omei Turnbull, Player Developer

  • 988 Posts
  • 310 Reply Likes
It's not a distraction; it's a legitimate bug report.  The browser should enforce its own requirements, not leave the user to figure out what they are.

Thank you for posting.