#1: Jan Wolter (jan) on Feb 9, 2009

So I've spent the last month or two doing a huge webpbn rewrite. I expect to be putting the new version up for beta test later this week (there will be a link that you can go to to use the new version of the site, but the old version will still be up), and then put it into production whenever I feel like it's been adequately tested.

When you look at the new site, I think you'll be amazed to see that it appears to be exactly the same. Yes, that's right. I rewrote the site to work pretty much exactly the same way it already does. Then I rewrote it again. And again. And again. Now it is much better on the inside, but still pretty much the same on the outside. They call this an "obsessive compulsive disorder". Oh well, there are some new features, and the inner infrastructure is much better for adding other new features.

So here's the list of changes, so far as I remember them. They are concentrated in the puzzle solving environment, with only a few minor changes on other parts of the site.

• More Javascript, Less Perl. When you asked to do a puzzle in the old environment, a Perl program would run on the server to generate a custom HTML page for you to solve that puzzle. The page would be different for every user and every puzzle. It would be customized for every browser. It would include a lot of Javascript code which was also customized for every browser. All this gets sent over to your computer, and then the Javascript program starts running.

  In the new site, there is one page "play.html" that you always load for any puzzle you want to solve. It loads an huge pile of Javascript, but it's always exactly the same Javascript for all users, for all browsers, for all puzzles. In practice, your browser will probably only load this once, and then get it from cache for all future puzzles you solve, just checking with webpbn to see if there is a newer version available. So this part should go very fast with very little work done by either your computer or the webpbn server.

  Once the Javascript application starts running, it makes three requests to webpbn.com to get your personal info (login name, full name, option settings, etc), information about the puzzle (size, clues, etc), and any saved info you have for the puzzle (saved games, ratings, etc). Then the Javascript does the drawing of the puzzle, and the server doesn't get involved again until you do a save.

  So the first advantage here is that it puts less load on the webpbn server, and more on the user's computers. Unless you have a really, really slow computer, your computer is faster than the 1/1000th slice of webpbn's server that is dedicated to you so this is a speed increase for most users, and a load decrease on the server, which enables us to support more users. The next generation of web browsers is going to be substantially faster at running Javascript (like 10 times faster in some cases) so this advantage is only getting better.

  The second advantage is that it gets me about halfway to being able to do a web-independent version of webpbn. With that you'd be able to set up webpbn so that when your laptop connects to the server it quietly uploads all the newest puzzles from webpbn into the browser cache. Then you could go someplace you don't have net connectivity, and be able to solve the cached puzzles, saving games and ratings to your cache. Next time you connect to the net, all your saved ratings and games would be synced to the webpbn server. Setting up your computer to do this would be a one-click thing. Google had done similar things with G-mail, and has made the core technology available to other sites via Google Gears. It basically lets sites like webpbn turn themselves into something more like an application on your computer, that makes connections over the internet but doesn't need them to operate, although it happens to run inside your browser.

  This required rewriting the Javascript to be browser independent, rewriting a lot of Perl code as Javascript code, and massive restructuring of everything.

• More Object Oriented Javascript. The old version of the puzzle solving environment used a lot of variables that were global to the page. If I tried to put two playable puzzles on one page, it wouldn't work because the two puzzles would interfere with each other, clicks on one would be marked on the other and so forth.

  I restructured the Javascript code to encapsulate everything in object structures. This makes each webpbn puzzle on a page much more of an encapsulated object that can be embedded in any page without interfering with other things. At some point in the future I might be able to allow people to do things like insert a few lines of HTML into their web pages, and have a webpbn puzzle appear there, kind of the way Amazon ads appear on the "Books" page on this site. Puzzle syndication. I'm not even sure yet if I want to do that, but it's nice to be closer to having the technology.

  Doing this meant changing every variable name, and every function call in the program.

• Separation of Logic and Drawing. Currently, puzzles are draw with HTML objects - we put up a big black image for the background, and drop lots of little white images on top of it to make the grid cells. It's not the only option. We could render puzzles using SVG instead. That's called "Scalable Vector Graphics". Drawing pictures by stacking up images is stupid. SVG lets do draw pictures more sensibly - draw a line, draw a circle, place some text, shade it, rotate it, etc. Firefox and Safari do pretty good SVG, but IE doesn't. (Microsoft decided that instead of implementing the standard, they'd invent something new, called "Silverlight" to do the same thing. Notice how the letters "SVG" appear in "Silverlight". Don't we just love Microsoft?)

  Anyway, if I wanted to do a version of the webpbn client that draws the puzzles with SVG (or even silverlight) instead of HTML, then I'd want almost all the same logic (when someone clicks here, this cell should be colored red; when they hit the save button we should save, etc), but different drawing. But in the old code the drawing code was all mixed into the logic. Now it's separated out. I even had a working SVG version of the puzzle client for a while, but later changes broke it. I'll need to get back to that later.

  So in the future this would let me sense if you are on a browser with good enough SVG support and render the puzzles that way instead. SVG would probably work faster and smoother. Scaling, for instance, is instantaneous, you don't have to redraw the whole puzzle bigger, instead you just change one thing that says how big to draw it the and browser instantly redraws it. No upper and lower limits for scaling either.

  All this, and all it required was another complete reorganization of the the program structure.

• Consistent Internal Representation of Colors. There are only five colors in webpbn, plus "unknown". You'd think I wouldn't need to have more than five different ways of representing them. But no, the old client, for historical reasons, was using different codes to represent different colors in different places. It was hopelessly confusing. To reduce my confusion I rewrote everything to use just one main encoding of colors. Of course, that meant every piece of code that had to do with colors had to change.

• Movable Clues. Finally, a change that actually gives us a new feature! I wanted to be able to move the clues. Why should they always be on the top and left side? Why not be able to move them to the bottom or the right? This would be very handy for people working on big puzzles - reduces scrolling. But to do it without a complete redraw of the grid, I need the clues to be all lumped together into a single grouping object, and have their positions defined relative to that object, so I can move them all by just moving the clues. So, the logical structure of how the elements on the screen are structured needs to change. Another big round of code changes.

• Tips. At the bottom of the screen, tips on now displayed. These give information about how to solve pbn puzzles and how to use the webpbn interface, including all the obscure features. You can cycle through the tips, or pull up specific ones from a menu. Experienced users who know all that crap can also turn them off.

• Reauthentication. Currently if your session expires while you are solving a puzzle, you may well be out of luck. The save won't work and you can't always relogin without losing the state of your puzzle. The new client should be smarter about this. If you session expires, it will ask you to relogin, and then it will do the save.

• Mouse Function Assignment. Currently there is a big box under the puzzle where you can assign different functions to different mouse buttons. I don't think a lot of people use this. I wanted to make it more compact, so now there is a single small button that controls that.

• Color Picker. I also added a new mouse button function. You can set your right mouse button (or SHIFT-click) to be a color picker. Click on a clue number, or a painted square with the color picker, and it sets the left mouse button to paint that color.

• Puzzle Options There is a new options button on the puzzle solving screen. It lets you save new setting either just for the current puzzle, or as defaults for all future puzzles. This includes turning the error checker on or off, assigning functions to mouse buttons, and turning off the tips.

• List and Find Pages. There have been much smaller changes to the puzzle list page. You can set options for a default list order, and filter out puzzles with multiple solutions. Find has a few new search and sorting options too.

That's all I can remember right now. There are some minor changes and bug fixes in other parts of the site, and since I ended up changing pretty much 100% of the puzzle solving client, lots of old bugs were fixed and new bugs were introduced.

I'm convinced this is going to end up as a much better setup, but I'm not convinced that it was actually worth the effort.

great news Jan can't wait to try this "no-change" change :)

I'm constantly amazed how much work you put into this. Thank you!

Jan, your generosity of time and talent is greatly appreciated. This site gives so many people a chance to escape the stress and anxiety of life and a chance to interact in a fun way with a pleasant community. You have done all the work, and we get the benefit. I hope that all you have done for us is returned to you one hundred fold in a good way down the road in the not-too-distant future.

Thanks.

Another change that will be in the new version:

When someone changes a published puzzle, creating a new version, webpbn will automatically post a message to the forum saying "Puzzle changed by joeblow".

This is to reduce confusion when people are commenting about earlier versions of a puzzle. I might also have it post messages like "Puzzle ruled logically solvable by jan", "Puzzle unpublished by jan" or "Puzzle republished by joeblow", but I don't want to do too much of that kind of stuff, because people reading new comments in the forums don't want to see nothing but a bunch of robot messages. But it would be nice to have the life history of the puzzle interleaved with the comments about the puzzle, and most puzzles don't have that much life history.

I've used the mouse function assignment since I started on this site. I was always bothered by the fact that I couldn't save my preferences though.

It sounds like your refactoring will be worth it just from the maintenance perspective of things. And hopefully the new code will make finding the bugs a little easier for you.

Saving mouse button preferences would be nice, but changing the options has become such a habit by now I barely notice. Personally, if I used the colour picker I'd have it on the middle button, since right-click is and always will be white for me.

In general, some neat looking stuff there, I look forward to the new version :)

OK, I've placed the development version on line for beta testing by anyone who wants to try it.

Notes:

• This is still quite buggy. Only courageous users ought to mess with it.

• It does not work with Internet Explorer. I've only just begun the long process of discovering and working around all of Microsoft's bugs, so that IE users can persist in the illusion that they are using a web browser and not a smoking cow turd. Someday web developers will stop doing this and Microsoft will be out of the browser business.

• Please report bugs in this item. This includes little tiny bugs, big nasty bugs and mere misfeatures. It does not, at this time, include IE bugs. I know about plenty of IE bugs, thanks.

• The beta site may occasionally stop working. This is mostly because I'm tearing its guts apart trying to fix something.

So to use the beta test version of webpbn, just go to:

http://webpbn.com/beta

it's says "404 not found"

Oops, typo in the link. Fixed now.

I'm using Firefox 3 on XP:
The 'Puzzle Play Options' are not saving properly. They load fine for the puzzle I'm on, but as soon as I click off and load a new one, it switches back to default. I tried both saving for the session only and setting as default.

A more minor issue is one that also bothers me on the current version, but I never got round to mentioning it. The middle mouse button sort of works, but the scroll thing always activates and messes it up (I forget the proper term, but where the cursor turns to some arrows and it scrolls just by moving it). I don't know if it's technically on your end, but it used to work fine, I'm pretty sure it broke when I upgraded from FF2 to 3.


Other than that, seems good so far :)

Oh hey, puzzles I solve on the beta still count on the current version! I shall definitely be poking around more ethen :D

About the scrolling/middle mouse button...I have the same problem. I just figured it was my mouse, but it used to work. Now the scrolling thing pops up and moves the puzzle and colors whatever the cursor rolls over without holding down the wheel button.

I don't know if I'm imagining this, but the click and drag motion seems to be smoother on the beta. It would tend to stick for me so that if I tried to click twice to black then drag a line, I would end up with a dot on the start and end squares. I got used to slowing down to prevent this from happening. It seems like the touch is "lighter" now, if that makes any sense!

Tip #17: Moving the Clues is SO COOL!

Oh yes, nearly forgot: Blue isn't loading correctly when picked with the color picker. It's fine otherwise.

How come everything is centered in this forum after Jan posted the link, or is that just my browser? And did you fix the search fuction in the new beta so that weekly challenge puzzles may be searched for? Just asking. I'm still using explorer (will migrate to firefox when I have more time) so I'm not able to check for myself ...

I tried 'WC' and 'WC#1' and got nothing.

The forum looks normal to me, but I'm guessing the center code just didn't end quite right on your end.

Jan, after I go to options and choose to not show tips and click to remember this option for all puzzles, it seems to forget next time I open a puzzle.

On the buttons arrangement, I prefer to have "Save" in the top or second position (like the original) and the new mouse button designator at the bottom.

I'm happy to be able to sort puzzles by number of colors. Thanks for that.

Any possibility of including the Description field in the search options?

The problem with the middle mouse button not working right is Firefox. Firefox refuses to release control of that button. Very annoying. Nothing I can do about it.

The thing that barbarann reports with the mouse buttons on the old site vs the new site actually sounds like IE vs any other browsers. There's a IE bug that can't be worked around that means drags can't be started on the second click of a double click. The new site will have the same problem with IE.

Thanks for all these bug reports. I will definitely act on them. But maybe not today. Go kiss a loved one.

Not a big deal, but I used the beta site today to solve a big puzzle, and I ran into a small "problem", that's not really a problem.

While I was solving, and I would go over to the numbers to grey out one I was done with, I would accidentally miss the number and the grid would jump. Can you add to the wish list a lock for the grid? It's a wish list item, only.... definitely not a need.

Yeah, I've had some accidental clue jumps too. I think I can reduce the likelihood of that substantially. Also I've wondered if the click-on-clues way of triggering the jump is the best idea. Maybe using the arrow keys to move them would be better. Except arrow keys are a pain - they send different codes on every browser, so recognizing them is hard.

I don't know how easy it would be to actually code, but I will say my first intinct upon trying to move the clues was to click and drag. Actually took me a couple attempts before I realised just a normal click was required >.>

So, I think http://webpbn.com/beta/ now works in IE 7 and IE 8. Still need to test it in older versions.

I made some changes so that the clues would be less likely to jump on errant clicks. I also made the arrow keys an alternate method of moving the clues.

I could probably implement a clue drag, but (a) it would be a bit complex to program, (2) it would feel strange if there was only one place you could drag them to, and (iii) it would be hard to do the drag when the puzzle is bigger than your screen, which is exactly the time when you'd most want to be able to move them.

OK-- I haven't been able to try it out yet because I usually solve these at work, and they don't trust me to download things, so I only have IE there. Finally, tried it out. No problems at all. It seemed exactly the same except the pages loaded a little faster, and I assume that's what you're going for! I did accidentally move the clue boxes, and that annoyed me for a while because I didn't know how I did it. But I found the tip for "moving clues" and figured it out. So, thumbs up from an IE user!

Oh-- that's what you guys were talking about with the "clue jumps." BTW, the arrow keys and the clicking both worked fine for that once I figured out what was going on.

I'm thinking of turning off the clicking and just supporting the arrow keys. They are less likely to be accidentally activated.

I'm a fan of using the H, V, and S keys. Would it be possible to allow the Z key to undo? Thanks!

That's not only possible, but easy. I just added it in.

woo hoo! thanks so much!

I booted up my old computer and started testing the new version with IE 5. Gawd, I'd forgotten how horrible that is. I beat down the first few bugs, and then hit one that crashed the browser every time I tried to solve a puzzle. That kind of thing is a bit hard to debug, so I decided it was time to reconsider.

Statistics indicate that usage of IE 5 has dropped below 0.5% of all users. IE 6 is at about 20%, IE 7 at 25%, IE 8 at 0.6%. (Firefox is at 44%). At least that's one source. Different people give very different statistics.

IE 5 doesn't work with google. When you click on a link on the google search page you get a browser error instead of going to the page.

So, my conclusion is that I can stop supporting IE 5.

But I still need to support IE 6. So I need to upgrade a windows 95 system from IE 5 to IE 6. So far, this is proving difficult. Microsoft doesn't seem to provide old versions of it's browsers. There are some available from other people on the internet. Dunno how reliable those are. Haven't been able to get one to install yet.

Oh well. Nothing but fun in the internet business.

Hmmm...windows 95 can't even unzip the install files for IE 6.0. Not a promising beginning.

Looks like IE 6 can't run on windows 95. I'll need to find another computer to test on.

At work, when we have tech questions that can be resolved by telling a person they should upgrade their browser or their adobe reader, that's just what we do! They're free programs that take just a few minutes to download. Are you being too nice by putting in such an effort for an outdated browser? I'm usually the first one to applaud an act of kindness, but in this case, isn't it the user's responsibility to keep somewhat current?

20% of the internet is still using IE 6, and some people have good reasons for not upgrading. Many work places won't let you upgrade anything until some ponderous IT department finally declares that everyone must upgrade.

IE 6 is still a fairly current browser, and I don't expect to have too hard a time getting it to work, now that I've got IE 7 and IE 8 both working. IE 6 and IE 7 differ a lot in their user interface (which is one reason some people don't want to upgrade), but they aren't very much different under the hood. In fact, it might already work. I just need to get my hands on a computer with IE 6 so I can test it.

I decide to load up IE6 and have a look. As far as I can tell, everything works fine. I solved a couple puzzles, tried moving the clues about and changed the size without issue. The only real difference is a slight change in how the site is displayed compared to Firefox 3, and even that is only noticeble because a horizontal scrollbar appears. In fact, I looked at the current homepage in IE6 as well and it looked identical to the beta, so it's obviously nothing important.

Also, when I paused the mouse over the grid lines, the picture toolbar appeared. Didn't cause any problems at all, it just amused me :)

I'm fairly certain I have IE6 (there are no tabs for multiple addresses) at my Tuesday/Thursday job, and I just solved a puzzle on it without any problem.

Ok-- I did find something very minor after all. After I posted my comment, there was no link for "go to next new topic." I had to click on the forum button on the left again. If I don't leave a comment, the "go to next new topic" link stays there.

Now that you mention it, that happened to me on Firefox. (#39)

yep, i had that issue last week... meant to mention....

Love, love, love being able to move the numbers! Thanks, Jan!

Sounds good on the IE 6 tests. I still haven't found a solution to that. Might need to buy an old second-hand computer so I can test on all versions of IE.

Yeah, I know about the "vanishing next topic link" bug. It's on my list. Probably an easy one to fix.

Most browsers have an option like "About Internet Explorer" on a pull-down someplace that will tell you what version number you are running.

So I'm thinking this is done. Well, not "done", "done" is a mythical state in the software lifecycle, but ready for release. Several more little features have been added in, and it seems to be working pretty well for me on all my many many browsers.

Probably I won't actually do the update till Monday though. It's never wise to do big software updates on Fridays.

So, I meant to be finished with this thing, but I accidentally did another big rewrite of big parts of it.

The error checker, which puts up the red-balls, and the helper, which tries to solve the puzzle when you click on the "helper" button, both used a feature of Javascript called "regular expressions". The FAQ discusses how these are used in some detail. However, my use of regular expressions is a little radical. My expressions are much longer and more complex than those used in typical applications.

Theoretically, that's not a problem. In practice, well, some browser developers don't seem to test their regular expression parsers on anything really complex. The opera browser used to choke on my regular expressions, but seems to have finally been fixed. Now the newest versions of Safari, which supposedly have radically faster regular expression checking code, turn out to have broken regular expression checking code, that just barfs when confronted my regular expressions.

So it occurred to me that I could take the line checker I wrote for the pbnsolve program and use it on the website instead of the regular expression stuff. So I restructured the Javascript code so different linesolvers/helpers could be plugged in, and ported over the pbnsolve linesolver.

So now there are two versions of the code. The new version is about 600 lines of code, compared to the old one which is about 300 lines of code.

Currently the beta site uses the new version for:

• Firefox 3 and later
• Safari 3 and later
• all versions of Opera

It's done for Opera and Safari because I don't trust their regular expression code. It's done for Firefox 3 because when I was testing I actually felt it was faster with the new code. The original reason I used regular expressions instead of writing a big Javascript program was mostly to make my life simple, but partly because Javascript was darned slow in those days. Firefox 3 has pretty fast Javascript, and the next generation of browsers (like Safari 4) will have even better.

So I haven't tested this nearly enough, but it is likely that the beta site will now work right with Safari.

Jan, I don't know if this is a PBN thing, or a my computer thing, but for the last couple of hours, the site keeps getting hung up on "Drawing puzzle" (or whatever it says) and the puzzle never gets drawn. Everything works fine in the regular site, but keeps getting hung up in beta....

Drat.

I can't tell much from the error logs, except that you are using the latest and greatest firefox on windows, which should be fine, and the last puzzle you solved on beta is #5244, where it looks like everything went OK as far as I can tell from the webserver side. Did this problem go away?

Are there specific puzzles you have trouble with? Any other patterns to when it works and when it doesn't?

Yeah. the problem is gone now... lemme go back and look to see if I can remember what puzzle specifically I noticed it first.

I think it was 5232. I tried a couple of others too... including ones I had previously solved and they all hung, so I just switched back to the old site. Tried the beta again later and it seemed to be working fine again, so I just stayed... wish I could help more. :\

This isn't a bug, I just noticed the middle mouse button appears to be working :D



EDIT: Never mind, false alarm. Must be one of those rare and mysterious 'beneficial' bugs :/

The middle mouse button does work on some browsers, but Firefox is possessive about it and won't let me override the default actions that they attach to it. However in some circumstances the default action doesn't amount to much and it can give the temporary illusion of working.

Something just occured to me while solving a wide puzzle: You can swap where the clue numbers are, but the count of how many squares are in a row/column is always on the opposite side to the clues, so there's still a bit of scrolling.
It's certainly not a major issue (since I could always just, you know, /count them myself/ >.> ) but still with everything else fixed so nicely, it's just one last obstacle to large puzzles.

i had some problem with the graying out like when you have finished a row or collum it grays all the numbers.

In the beginning it grayed out the row and collum if i draged the mouse paining a few black or "white". Then i swithed the number-clues to the other side and i stopped that even after swithing back but then it didn't gray out when i did finish a row or collum not even at the end when i finish the puzzle.

I hope its an easy fix since i really like the grayng out fiture.

But i love all the rest of it, specially moving the numbers.

I'll look into it.

So I just finished installing the new version of the software on the site. So in the extremely likely event that there are still lots more bugs hiding, I guess I'll soon be hearing about them. This kind of massive rewrite/reorganization of the internals of the site is bound to cause trouble.

im no longer having any problems.

It saves fast it loads fast. Love the new find puzzles options. Love that you can go back to your search and pick the next one.

LOVE love Love LOVE everything Jans the man :-)

I am having a problem when occasionally the list of numbers at the top of the grid suddenly flip to the bottom of the grid, or left moves to right. Also sometimes the numbers stay at the top and seem to move to be justified along the top line so you get columns of numbers like this

1 8 11 3 9 6
2 1 1 2
4 2 1 1
2 4
1
1
1

That's supposed to be a feature. The arrow keys move the clues from side to side and also switch them between left and right justified. You can also move them by clicking on the white space in the clue area.

Moving them to different sides of the puzzle is useful on large puzzles that don't entirely fit on your screen. Changing the justification makes it easier to look at the clues relevant to one edge, or to grey them out. In your example above, if I was working on the left edge of the puzzle, this justification might be easier to look at than the normal one. Or if I had filled in a lot of stuff on the left side of the puzzle, it would be easier for me to gray out the clue numbers with a click and drag over them.

I think the click-on-white-space-in-clue-area way of moving the clues tends to cause accidental clue moves, which are confusing. I'm thinking of turning that off and just making them movable by the arrow keys, or by options in the option menu.

I agree with your thinking. It's very easy to accidentally click in the white space when trying to gray out a number. The arrow keys method (didn't realize that was another way to do it) sounds much less prone to accidental clue swapping. I'll have to play with the justification - definitely would make it much quicker when graying out clues on the top or left.

OK, click on the white space around clues no longer moves the clue numbers. Now the arrow keys are the only way to move the clues.

Jan, I'm having trouble with the Helper. When editing a puzzle, I go to the solve page, and hit the Helper button, then it says "solving..." but nothing happens after that, the blank grid just sits there. I can continue to solve it myself, and save, but if I hit the Helper button again, it still hangs with the "solving..." message and no action. I opened other puzzles I have made that are unique and solvable and the same thing happens. I am using Firefox 3.0.7. I closed out Mozilla and tried it with IE and had the same experience. I really like the Helper feature. Please help!

I'll check into it.

The helper seems to be working for me, so I'm going to need a bit more help from you to figure this out.

The helper has always been a separate program that is loaded when you click on the "helper" button. My suspicion is that there is a problem loading that file. So let's try a few tests.

The helper is loaded from http://webpbn.com/jsz/helper_reg.js

Clicking on that link should give you a text file that begins with my copyright message. If you get an error message or a page of gobblity-gook, then that might give me an idea of what your problem is.

It occurs to me that you make a lot of puzzles, and must use the helper often. Did this problem start in just the last few days? I haven't made any change to that code in a week or so. Has there been any change with the computer you are using?

The text file opened just fine, no problem there.

About a week ago was when it first happened. I thought it was just my computer having a bad day, and so I waited to see if it happened again after I restarted my computer. Still hanging.

If this were "Car Talk" I'd be joking about how you just gave me the worst possible answers, completely spoiling all my favorite theories.

Now I need to think up a new theory. Hmmm....

I haven't exactly got a new theory, but I have something new to try. Some kind of Javascript error in the helper program could cause the behavior you describe, lets go looking for a Javascript error message.

In firefox go into one of your own puzzles. Before clicking the "helper" button, go to firefox's "Tools" menu and select "Error Console". This should pop up a new window. It may have some error messages left in it from previous web pages. Click "Clear" to get rid of them. Then, leaving the error console window open, hit the "helper" button on the main screen. I'm interested in hearing if any error messages appear on the error console.

I followed your instructions. There was an error listed for the helper reg.js:

fm(fi) is undefined line 84

I cleared it, then clicked the Helper button, and it still hangs. The error console shows the same message again. I cleared it again and tried a different puzzle. Still hangs. Hopefully the error message will give you a clue.

Another error is showing up in the error console along with the helper error message:

PBN.makeTwoRE is not a function
http://webpbn.com/jsz/playerhr.js line 1802

I had someone else log on from another computer and try it. They had the same problem with it hanging with the "solving..." but the error message for helper reg.js is different:

pbn.line.er(d) is undefined line 50

OK. The PBN.makeTwoRE error message is the key one. I know what's wrong now. I just need to fix it. This is a little bit of code that doesn't get exercised for me because I am an admin. It probably effects all non-admins.

Now I just need to fix it.

OK, I think I fixed it. Lots of thanks to Teresa for all the testing and reporting.

Yeah - helper is back! Works fine, right away. No errors in the console. Thanks so much, Jan. Now I can make more challenging puzzles!

I've been getting an "unresponsive script" error message for this script: http://webpbn.com/jsz/playhr.js:1280 The message says "A script on this page may be busy, or it may have stopped responding. You can stop the script now, or you can continue to see if the script will complete."


It is only happening while I am trying to solve puzzle #5449, even after starting over 3 times. It is most likely to happen while I am doing a click-and-drag to fill in a series, but it has also happened while just clicking to fill in one pixel. I just tried using the webpbn.com/old site and the same thing happened, while doing a click-and-drag fill. I have been using Firefox, but I did try with IE, where I don't get the error message, but the response time for clicking through colors is very slow at times.

Regarding the problem I posted in the previous comment, there are no error messages in the error console under "tools."

having the same issue with the same puzzle. Latest firefox.

I had this problem yesterday with Firefox on a different, very large puzzle. It was working fine until I tried to use h or v to fill a line, but then it froze and I had to close the browser. I thought it might have been due to a routine scan running in the background.

I was asked to mention that on "Puzzle #5449: WC #29 - The best board game ever" some have received error messages including me while working on the puzzle. With me it completely froze my IE. I know this second to last time it happened as soon as I tried to resize the puzzle. But the last time I didn't resize it and it never froze up on me.

The problem is caused by the error checking script on large puzzles. Click on options and turn off error checking and you shouldn't freeze on the puzzle.

Robert is probably right - we've always had problems with the error checking code becoming extremely slow on large puzzles.

I've written a new version of the error checking code that is currently used on Safari and Opera, It isn't faster than the old version, in fact it usually seems slower, but it avoids certain bugs in Safari and (maybe only older versions) of Opera. However, though it isn't faster, I could make it time out cleanly, so that if an error check is taking too long it doesn't freeze the browser. So maybe there will be a way to "solve" this problem eventually, but it's going to take some doing. Meanwhile, it's often best to turn off error checking on large puzzles with lots of clues.

Jan, I appreciate your working on the code for the error checker. I am one of those who have become dependent on the error checker, especially for the large color puzzles where I cannot detect the color of the number clues.

One more thing: A while back, I asked if you could include the Description field in the search page, and I think you said that was possible or a good idea or something like that. I am still hoping you can do that. There is a puzzle I am looking for, with a specific image, but the word for the image is not in the title and is not in the comments, is only in the description, and so I cannot find it.

Thank you, Jan, for maintaining this site and for helping us out with all our requests.

I'm just back from a week long vacation. I was theoretically supposed to have internet access from the place I was staying, but it never materialized. I hope to find some time to do some bug fixing this week, but I'm not planning a lot of new feature stuff in the immediate future. Need time for other projects.

I had some trouble after loading a saved puzzle which solution had changed when i hit save after having solved alot of the puzzle it just filed the rest in with blanc (the dot) and said that i had solved an old version of the puzzle. It was on puzzle #4779: Not another camel ...

It happent twice in the same puzzle. im guessing it will very rarely happen so its not that importent to fix but though you should know anyway.

I had the same problem recently. I was redoing some previously saved puzzles that had changed, and when I hit save, it filled in all the blanks with dots. Happened on two different puzzles.

Goto next topic

Users cannot post to this thread.