Like everyone else I am having ongoing issues with the posting editor. It will hang at odd times and it has nothing to do with "check your settings etc" as it clearly has to do with the remote server of which I am absolutely sure. I know how to check this and that and I also know that there is no real cpu load this end either. I have wireshark running continually in the background just to spot where it hangs and what it is waiting on. I even closed firefox on my Linux Mint 20.1 system and created a debug profile just to test it further.
Mind you, I tested this about 1pm AEST with my default profile and it was fine. Of course it would be fine, even though I was copy and pasting an edit of one of my posts, because I was testing it with wireshark, and it knew it But this could be related to time of day since earlier on it was hanging for long periods, but I didn't have wireshark running, so it knew it could misbehave!
Now I keep typing but it is not playing up, so let me attach some files and an image for a test, but please ignore these for the test.!
Perhaps I will try to keep track of when this is happening though, It is fine at around 3:30 UTC but I am sure that around midnight UTC I had problems. Has anyone noticed if it has played up at certain times more than others?
That's funny, I attached an image of SED and the first time it didn't show. Then I attached it again and it shows although the edit of the post clearly shows the line for the first image as well, although only the second one shows.
EDIT: somehow the ! went missing at the start of the link. Anyway, I removed it after testing it.
In case this hasn't been highlighted-
Time was when I could lookup all the replies/comments I have made on the forum over the years. Now we have the item 'participated' which does NOT do the same thing. It only takes you to the END of a topic in which I had at sometime participated, not to the actual item itself, which might be hidden among hundreds of other comments. To be clear this is not for 'My Discussions' which I have started, but to item like this item I am writing now as part of another topic.
Is there any prospect of this functionality being restored or am I using the wrong facility.
@tritonium
If you click on your own user name, then look to the left hand of the screen, you should see menu links for "Discussions" and "Comments" that comprise all your contributions.
@evanh said:
Kind of begs the question of how the older forum software achieved a larger edit box that worked with older browsers.
I think the old box was just fit to text height but with a minimum. For some reason this one adjusts its height from a script (and it doesn't even work properly. Try previewing a post with lots of images and then going back to edit mode....)
@evanh said:
Kind of begs the question of how the older forum software achieved a larger edit box that worked with older browsers.
I think the old box was just fit to text height but with a minimum. For some reason this one adjusts its height from a script (and it doesn't even work properly. Try previewing a post with lots of images and then going back to edit mode....)
Old edit box was at least ~250 pixels high, probably, much taller than now and however the height was set it worked fine with a 10-year browser.
@Wuerfel_21 said:
I'm not super familiar with it, either, but what you're looking for is the "Main" section. Also, there's buttons to load and save profile captures next to the record button. Next to that is a dropdown where you can view the recent captures.
Here's the hanging capture. Note how enormously long those tasks are and the "estimated blocking time" being astronomical.
The calls to T take multiple seconds each.
Annoyingly, the scripts are all obsfuscated, so it's somewhat difficult to tell what exactly is hanging. If the server had the relevant source maps, it'd be easier.
The performance monitor is showing a couple of details that are troubling. The first is the almost constant execution of Microtasks. This is the Javascript engine processing asynchronous calls. If you look at the call tree, you can see that whatever is running is iterating through an array of objects over and over again. The high idle time is possibly indicating that the async calls are accessing a network resource with some latency.
The other issue is that the performance monitor is indicating that heap memory balloons significantly while this activity is running. In the graph, it starts at about 2200ms.
I will try to reproduce this locally and figure out what is doing all of the bit-banging.
No offense intended, but there are so many links to images, urls, old posts, etc, that are broken,
I would BEG any new consideration to change the forum yet again, to please stop.
Even old links still haven't been parsed and their urls corrected to the new forum post path. From 2 previous forum changeovers ago.. those links are still dead.
I find sooo many when trying to search it for help, I find the help, and the link the user posts to the forum page that has the help....
IS 404.
Nice pic btw, parallax. I kinda feel like how that kid looks when I search for info and encounter yet another dead link to a parallax forum post.
'end rant.
Really I just wanted to repost that image, so funny.
How difficult would it be to set up a series of automatic redirects for forums.parallax.com to listen for old urls and redirect to new versions of them, instead of crawling the forums and replacing old links? It would make the process of fixing links much faster as far as I understand.
@AwesomeCronk said:
How difficult would it be to set up a series of automatic redirects for forums.parallax.com to listen for old urls and redirect to new versions of them, instead of crawling the forums and replacing old links? It would make the process of fixing links much faster as far as I understand.
The listening part is not difficult. The detecting and redirecting are difficult. There are at least three different URL constructs used on this site over the years since vBulletin. To do this well, we would have to identify a distinct pattern in each URL and then look up the current reference to the old URL. Which means that we need something like a dictionary to map the old to the new. It's complicated.
@AwesomeCronk said:
How difficult would it be to set up a series of automatic redirects for forums.parallax.com to listen for old urls and redirect to new versions of them, instead of crawling the forums and replacing old links? It would make the process of fixing links much faster as far as I understand.
The listening part is not difficult. The detecting and redirecting are difficult. There are at least three different URL constructs used on this site over the years since vBulletin. To do this well, we would have to identify a distinct pattern in each URL and then look up the current reference to the old URL. Which means that we need something like a dictionary to map the old to the new. It's complicated.
Hello!
Jim? How complicated? Was the Old Forum kept available? If I remember correctly during the first revision, the Old Forum was kept up by sticking a notation on it that insists it was in Read-Only mode and we could only read those posts, and everything new was going to the New Forum in its next series of updates.
And the mascot is away for the coming week, he's visiting family.
Um AC? Be careful, you're about to be invaded by a small army of robots who walked all over the Mojave to find a better place to annoy.
@Cluso99 said:
I thought Phil (@phipi) wrote a conversion program years back that went thru the old links and converted them. Cannot this be done again???
For sure. One approach would be to run an SQL script to fix most of them up. And a little helper/resolver program to catch the more obfuscated ones.
I think it's not that difficult. For at least one of the link types, it's just the same thread id in a different configuration, that's a simple regex operation.
If y'all can list all the different URL patterns, it'd be trivial to write a script to to do it client-side for now.
So yeah, a few moments thought and an sql query. In fairness, there's a whole bunch of more cryptic styles too which may need a little helper to search & resolve. I don't have an example with me, but I think we've all seen them. A quick google search is the way we usually fix them up manually, so it's certainly not beyond the efforts of a script if the db doesn't provide a smarter way.
// ==UserScript==
// @name Parallax Link Fixerator
// @version 0.1
// @author Wuerfel_21
// @match *://forums.parallax.com/*
// @run-at document-end
// ==/UserScript==
(function() {
'use strict';
const links = document.getElementsByTagName('a');
for (const lnk of links) {
console.log(`lnk is ${lnk}`);
var url = lnk.getAttribute("href");
if (url == null) {
// dum dum dum dum dum dum dum, du du dum
} else if (url.match(/^https?:\/\/forums\.parallax\.com\/showthread\.php\?(\d+)-([\w-]+)/)) {
console.log("MATCHED!")
lnk.setAttribute("href",`https://forums.parallax.com/discussion/${RegExp.$1}/${RegExp.$2}`)
}
}
})();
I thought Phil (@phipi) wrote a conversion program years back that went thru the old posts and converted them.
Yup. Back in 2010, I wrote a Perl script that logged in with admin privileges and spidered the entire forum doing search-and-replace using Perl's regex facility. Once it was proven on a forum copy, Jim ran it late at night on the real forum after general access was blocked. It took several hours to run.
Hello!
And now there's something completely different to report. I use Chrome (Not the Microsoft branded one.), and selected here from a selection of locations that the browser decided I visit most. It then showed me a "Site took to long to respond error message", and chose to translate HTTP 504 as that. The site came up quickly when I pressed the reload button.
All in all that's preferable to a HTTP 500 one, which means the site server itself crashed and is waiting for someone or indeed something else to restart it.
Mascot on vacation.
Now I wonder why there is a bigfoot convention in the parking lot of Parallax.
Hello!
Okay next one. How to I enclose code blocks in these proceedings? The select commands in PBASIC use the hash mark to cause the Stamp to do so. But in Markdown it causes the text to become bold ones.
There was some discussion earlier in this thread to do so, but, ah, I can not seem to find it.
' =========================================================================
'
' File...... screentest8.bs2
' Purpose... screener
' Author.... GCL -- Jedi Knight Computers
' E-mail.... gregg@levine.name
' Started... 10 Mar 2014
' Updated... 11 Mar 2014
'
' {$STAMP BS2}
' {$PBASIC 2.5}
'
' =========================================================================
' -----[ Program Description ]---------------------------------------------
' This program outputs video to an NTSC color monitor via a Propeller
' Backpack module and demonstrates the various features of the video driver.
' -----[ I/O Definitions ]-------------------------------------------------
io PIN 15 ' Serial I/O pin for Propeller Backpack.
U1A PIN 0
U1B PIN 1
U1C PIN 2
' -----[ Constants ]-------------------------------------------------------
' Baudrate definitions. Serial I/O must be open-drain. The Propeller Backpack
' includes a pullup internally.
#SELECT $STAMP
#CASE BS2, BS2E, BS2PE
baud CON 84 + 32768
#CASE BS2SX, BS2P
baud CON 240 + 32768
#CASE BS2PX
baud CON 396 + 32768
#ENDSELECT
' The following is a table of "command" constants for the Propeller Backpack
' that can be copied to other programs. The ones that are commented are already
' defined by the BASIC Stamp Editor and perform the same functions as they
' would in a DEBUG screen.
'CLS CON $00 'clear screen
'HOME CON $01 'home
'CRSRXY CON $02 'set X, Y position (X, Y follow)
'CRSRLF CON $03 'cursor left
'CRSRRT CON $04 'cursor right
'CRSRUP CON $05 'cursor up
'CRSRDN CON $06 'cursor dn
USECLR CON $07 'use color C (C follows)
BS CON $08 'backspace
'TAB CON $09 'tab (8 spaces per)
'LF CON $0A 'linefeed
'CLREOL CON $0B 'clear to end of line
'CLRDN CON $0C 'clear down (to end of window)
'CR CON $0D 'return
'CRSRX CON $0E 'set X position (X follows)
'CRSRY CON $0F 'set Y position (Y follows)
DEFWIN CON $10 'define window W (W, Left, Top, nCols, nRows follow)
USEWIN CON $11 'use window W (W follows)
DEFCLR CON $12 'define color C (C, FG, BG follow)
SCRLLF CON $13 'scroll window left
SCRLRT CON $14 'scroll window right
SCRLUP CON $15 'scroll window up
SCRLDN CON $16 'scroll window down
CHGCLR CON $17 'change all colors in window to C (C follows)
SCRSIZ CON $1D 'set screen size (Rows, Columns follow)
CLRW CON $1E 'same as CLR, but can be used in strings.
ESC CON $1F 'escape next character C (i.e. print as-is) (C follows)
ZERO CON $FF 'can be used for 0, which is not allowed in strings, for command arguments
'-------[ Variables ]----------------------------------------------------------
i VAR Word
char VAR Byte
'U VAR Word
U1 VAR Word
W VAR Word
'-------[ Program starts here. ]-----------------------------------------------
LOW io 'Reset the Propeller Backpack
PAUSE 500
INPUT io
PAUSE 2000 'Wait for it to come out of reset.
'SEROUT io, baud, [DEFWIN, 1, 20, 2, 9, 18, USEWIN, 1, CHGCLR, 12]
'SEROUT io, baud, [DEFCLR, 12, $AD, $0C]
SEROUT io, baud, [DEFWIN, 1, 20, 2, 7, 9, USEWIN, 2, CHGCLR, 13]
SEROUT io, baud, [DEFCLR, 12, $AD, $0C]
'FOR U = 65 TO 127
'FOR W = 1000 TO 1050
FOR W = 65 TO 127
'FOR W = 65 TO 91
'FOR W= 65 TO 99
'FOR W = 65 TO 255
SEROUT U1B, baud,[W] 'PIN1
PULSOUT U1A, 4 'PIN0
'PULSIN U1C, 8, U1
COUNT U1C, 4, U1 'PIN2
SEROUT io, baud, ["___>", DEC U1, CR]
PAUSE 1000
SEROUT io, baud, [" ", ASC ?W,CR]
PAUSE 350
SEROUT io, baud, [" ", DEC W,CR]
PAUSE 700
NEXT
'NEXT
Comments
Like everyone else I am having ongoing issues with the posting editor. It will hang at odd times and it has nothing to do with "check your settings etc" as it clearly has to do with the remote server of which I am absolutely sure. I know how to check this and that and I also know that there is no real cpu load this end either. I have wireshark running continually in the background just to spot where it hangs and what it is waiting on. I even closed firefox on my Linux Mint 20.1 system and created a debug profile just to test it further.
Mind you, I tested this about 1pm AEST with my default profile and it was fine. Of course it would be fine, even though I was copy and pasting an edit of one of my posts, because I was testing it with wireshark, and it knew it
But this could be related to time of day since earlier on it was hanging for long periods, but I didn't have wireshark running, so it knew it could misbehave!
Now I keep typing but it is not playing up, so let me attach some files and an image for a test, but please ignore these for the test.!
Perhaps I will try to keep track of when this is happening though, It is fine at around 3:30 UTC but I am sure that around midnight UTC I had problems. Has anyone noticed if it has played up at certain times more than others?
That's funny, I attached an image of SED and the first time it didn't show. Then I attached it again and it shows although the edit of the post clearly shows the line for the first image as well, although only the second one shows.
EDIT: somehow the ! went missing at the start of the link. Anyway, I removed it after testing it.
@"Buck Rogers" An asterisk at the beginning of a line is markdown for a bulleted list, for example, this input:
produces this result:
If you want any of the markdown characters to appear as plaintext, simply put a backslash before said character.
This is an asterisk: *
Still having issues just typing plan text. I can hear my computer just run ramped as the java script is looping like crazy.
Have we gone done a rabbit hole that doesn't give us support for this new fancy tool....
Mike
What changed because I am getting "long-running script" errors in IE11 and I usually only get those on the Yahoo main page?
Hi
In case this hasn't been highlighted-
Time was when I could lookup all the replies/comments I have made on the forum over the years. Now we have the item 'participated' which does NOT do the same thing. It only takes you to the END of a topic in which I had at sometime participated, not to the actual item itself, which might be hidden among hundreds of other comments. To be clear this is not for 'My Discussions' which I have started, but to item like this item I am writing now as part of another topic.
Is there any prospect of this functionality being restored or am I using the wrong facility.
Dave
@tritonium
If you click on your own user name, then look to the left hand of the screen, you should see menu links for "Discussions" and "Comments" that comprise all your contributions.
Edit box looks fine with recent Chromium but is only two lines high with an older browser (FF3.6.28) which means lots of vertical scrolling.
Any chance of making the edit box taller?
Perhaps Wuerfel_21 who is clearly a HTML wizard could take a look.
The height of the text area is updated by some script. However, it has a
min-heightof 250px that it should always maintain.In other words, old browser bugs?
Kind of begs the question of how the older forum software achieved a larger edit box that worked with older browsers.
I think the old box was just fit to text height but with a minimum. For some reason this one adjusts its height from a script (and it doesn't even work properly. Try previewing a post with lots of images and then going back to edit mode....)
Old edit box was at least ~250 pixels high, probably, much taller than now and however the height was set it worked fine with a 10-year browser.
The performance monitor is showing a couple of details that are troubling. The first is the almost constant execution of Microtasks. This is the Javascript engine processing asynchronous calls. If you look at the call tree, you can see that whatever is running is iterating through an array of objects over and over again. The high idle time is possibly indicating that the async calls are accessing a network resource with some latency.
The other issue is that the performance monitor is indicating that heap memory balloons significantly while this activity is running. In the graph, it starts at about 2200ms.
I will try to reproduce this locally and figure out what is doing all of the bit-banging.
The new layout of the forum screen didn't passed unnoticed...
Much more clear and informative, with a wider "search box", right at the top, and the moving images at the bottom.
Way to go!
No offense intended, but there are so many links to images, urls, old posts, etc, that are broken,
I would BEG any new consideration to change the forum yet again, to please stop.
Even old links still haven't been parsed and their urls corrected to the new forum post path. From 2 previous forum changeovers ago.. those links are still dead.
I find sooo many when trying to search it for help, I find the help, and the link the user posts to the forum page that has the help....
IS 404.
Nice pic btw, parallax. I kinda feel like how that kid looks when I search for info and encounter yet another dead link to a parallax forum post.
'end rant.
Really I just wanted to repost that image, so funny.
I’m still seeing unacceptable stalls when editing/adding a post.
I agree it's a pain sometimes.
More often than sometimes. Agreed that it’s not as regular, but still an unacceptable PITA
FWIW, if I wasn’t committed to P1 & P2 I’d go elsewhere. So for a newbie we might lose them before we get a chance to convert them!
How difficult would it be to set up a series of automatic redirects for forums.parallax.com to listen for old urls and redirect to new versions of them, instead of crawling the forums and replacing old links? It would make the process of fixing links much faster as far as I understand.
The listening part is not difficult. The detecting and redirecting are difficult. There are at least three different URL constructs used on this site over the years since vBulletin. To do this well, we would have to identify a distinct pattern in each URL and then look up the current reference to the old URL. Which means that we need something like a dictionary to map the old to the new. It's complicated.
Hello!
Jim? How complicated? Was the Old Forum kept available? If I remember correctly during the first revision, the Old Forum was kept up by sticking a notation on it that insists it was in Read-Only mode and we could only read those posts, and everything new was going to the New Forum in its next series of updates.
And the mascot is away for the coming week, he's visiting family.
Um AC? Be careful, you're about to be invaded by a small army of robots who walked all over the Mojave to find a better place to annoy.
I thought Phil (@phipi) wrote a conversion program years back that went thru the old links and converted them. Cannot this be done again???
For sure. One approach would be to run an SQL script to fix most of them up. And a little helper/resolver program to catch the more obfuscated ones.
@Publison and me fix up the links manually as we come across them, and it remains #20 on the list of tasks for the "global" solution : https://forums.parallax.com/discussion/comment/1515219/#Comment_1515219
I think it's not that difficult. For at least one of the link types, it's just the same thread id in a different configuration, that's a simple regex operation.
If y'all can list all the different URL patterns, it'd be trivial to write a script to to do it client-side for now.
I figured reflex would be involved. Thanks for getting back to me on that everyone!
Indeed. It's not difficult at all. I don't think anyone said that.
The simplest ones to catch are like....:
http://forums.parallax.com/showthread.php?90707-Propeller-development-for-non-Windows-users
Should be this: https://forums.parallax.com/discussion/90707/propeller-development-for-non-windows-users
So yeah, a few moments thought and an sql query. In fairness, there's a whole bunch of more cryptic styles too which may need a little helper to search & resolve. I don't have an example with me, but I think we've all seen them. A quick google search is the way we usually fix them up manually, so it's certainly not beyond the efforts of a script if the db doesn't provide a smarter way.
Yea, here's a userscript that catches those:
// ==UserScript== // @name Parallax Link Fixerator // @version 0.1 // @author Wuerfel_21 // @match *://forums.parallax.com/* // @run-at document-end // ==/UserScript== (function() { 'use strict'; const links = document.getElementsByTagName('a'); for (const lnk of links) { console.log(`lnk is ${lnk}`); var url = lnk.getAttribute("href"); if (url == null) { // dum dum dum dum dum dum dum, du du dum } else if (url.match(/^https?:\/\/forums\.parallax\.com\/showthread\.php\?(\d+)-([\w-]+)/)) { console.log("MATCHED!") lnk.setAttribute("href",`https://forums.parallax.com/discussion/${RegExp.$1}/${RegExp.$2}`) } } })();Yup. Back in 2010, I wrote a Perl script that logged in with admin privileges and spidered the entire forum doing search-and-replace using Perl's regex facility. Once it was proven on a forum copy, Jim ran it late at night on the real forum after general access was blocked. It took several hours to run.
-Phil
Hello!
And now there's something completely different to report. I use Chrome (Not the Microsoft branded one.), and selected here from a selection of locations that the browser decided I visit most. It then showed me a "Site took to long to respond error message", and chose to translate HTTP 504 as that. The site came up quickly when I pressed the reload button.
All in all that's preferable to a HTTP 500 one, which means the site server itself crashed and is waiting for someone or indeed something else to restart it.
Mascot on vacation.
Now I wonder why there is a bigfoot convention in the parking lot of Parallax.
Hello!
Okay next one. How to I enclose code blocks in these proceedings? The select commands in PBASIC use the hash mark to cause the Stamp to do so. But in Markdown it causes the text to become bold ones.
There was some discussion earlier in this thread to do so, but, ah, I can not seem to find it.
Mascot perplexed.
' ========================================================================= ' ' File...... screentest8.bs2 ' Purpose... screener ' Author.... GCL -- Jedi Knight Computers ' E-mail.... gregg@levine.name ' Started... 10 Mar 2014 ' Updated... 11 Mar 2014 ' ' {$STAMP BS2} ' {$PBASIC 2.5} ' ' ========================================================================= ' -----[ Program Description ]--------------------------------------------- ' This program outputs video to an NTSC color monitor via a Propeller ' Backpack module and demonstrates the various features of the video driver. ' -----[ I/O Definitions ]------------------------------------------------- io PIN 15 ' Serial I/O pin for Propeller Backpack. U1A PIN 0 U1B PIN 1 U1C PIN 2 ' -----[ Constants ]------------------------------------------------------- ' Baudrate definitions. Serial I/O must be open-drain. The Propeller Backpack ' includes a pullup internally. #SELECT $STAMP #CASE BS2, BS2E, BS2PE baud CON 84 + 32768 #CASE BS2SX, BS2P baud CON 240 + 32768 #CASE BS2PX baud CON 396 + 32768 #ENDSELECT ' The following is a table of "command" constants for the Propeller Backpack ' that can be copied to other programs. The ones that are commented are already ' defined by the BASIC Stamp Editor and perform the same functions as they ' would in a DEBUG screen. 'CLS CON $00 'clear screen 'HOME CON $01 'home 'CRSRXY CON $02 'set X, Y position (X, Y follow) 'CRSRLF CON $03 'cursor left 'CRSRRT CON $04 'cursor right 'CRSRUP CON $05 'cursor up 'CRSRDN CON $06 'cursor dn USECLR CON $07 'use color C (C follows) BS CON $08 'backspace 'TAB CON $09 'tab (8 spaces per) 'LF CON $0A 'linefeed 'CLREOL CON $0B 'clear to end of line 'CLRDN CON $0C 'clear down (to end of window) 'CR CON $0D 'return 'CRSRX CON $0E 'set X position (X follows) 'CRSRY CON $0F 'set Y position (Y follows) DEFWIN CON $10 'define window W (W, Left, Top, nCols, nRows follow) USEWIN CON $11 'use window W (W follows) DEFCLR CON $12 'define color C (C, FG, BG follow) SCRLLF CON $13 'scroll window left SCRLRT CON $14 'scroll window right SCRLUP CON $15 'scroll window up SCRLDN CON $16 'scroll window down CHGCLR CON $17 'change all colors in window to C (C follows) SCRSIZ CON $1D 'set screen size (Rows, Columns follow) CLRW CON $1E 'same as CLR, but can be used in strings. ESC CON $1F 'escape next character C (i.e. print as-is) (C follows) ZERO CON $FF 'can be used for 0, which is not allowed in strings, for command arguments '-------[ Variables ]---------------------------------------------------------- i VAR Word char VAR Byte 'U VAR Word U1 VAR Word W VAR Word '-------[ Program starts here. ]----------------------------------------------- LOW io 'Reset the Propeller Backpack PAUSE 500 INPUT io PAUSE 2000 'Wait for it to come out of reset. 'SEROUT io, baud, [DEFWIN, 1, 20, 2, 9, 18, USEWIN, 1, CHGCLR, 12] 'SEROUT io, baud, [DEFCLR, 12, $AD, $0C] SEROUT io, baud, [DEFWIN, 1, 20, 2, 7, 9, USEWIN, 2, CHGCLR, 13] SEROUT io, baud, [DEFCLR, 12, $AD, $0C] 'FOR U = 65 TO 127 'FOR W = 1000 TO 1050 FOR W = 65 TO 127 'FOR W = 65 TO 91 'FOR W= 65 TO 99 'FOR W = 65 TO 255 SEROUT U1B, baud,[W] 'PIN1 PULSOUT U1A, 4 'PIN0 'PULSIN U1C, 8, U1 COUNT U1C, 4, U1 'PIN2 SEROUT io, baud, ["___>", DEC U1, CR] PAUSE 1000 SEROUT io, baud, [" ", ASC ?W,CR] PAUSE 350 SEROUT io, baud, [" ", DEC W,CR] PAUSE 700 NEXT 'NEXT