Joined: 05 Oct 2008 Posts: 666 Location: Helsinki, Finland / nr. Alkmaar, Netherlands
Posted: Wed Sep 28, 2022 12:22 Post subject:
Upgraded R7800 to 50274 from previous f/w version using easyddup.
I have done a total reset on the previous f/w version and a manual restore mainly from PDF printouts.
BUG:
Sorting of the two last columns (service name and state) of the active connections table does not work in f/w 50274 after opening the table from the link in the status router page. Other columns can be sorted just fine.
After a refresh of the table using the button all columns can be sorted again.
Joined: 31 Jul 2021 Posts: 2146 Location: All over YOUR webs
Posted: Wed Sep 28, 2022 12:38 Post subject:
ArjenR49 wrote:
Upgraded R7800 to 50274 from previous f/w version using easyddup.
I have done a total reset on the previous f/w version and a manual restore mainly from PDF printouts.
BUG:
Sorting of the two last columns (service name and state) of the active connections table does not work in f/w 50274 after opening the table from the link in the status router page. Other columns can be sorted just fine.
After a refresh of the table using the button all columns can be sorted again.
Looks like a cache issue, old code cached headbutting new code, nothing to solve.
If the issues are gone after clearly cache which is what happened when you opened the page standalone so new code was pulled in, or pressing CTRL + F5 no point reporting them, only if they persist after clearing cache.
Joined: 31 Jul 2021 Posts: 2146 Location: All over YOUR webs
Posted: Wed Sep 28, 2022 14:09 Post subject:
The problem is that for some reason adding <thead><tr><th> for some reason breaks some tables, but works with the wrong <tbody><tr><th>, I think it could also be fixed by shunting the button to <tfoot> like I done in the static leases table to stop the button being sorted.
So really thead/tbody is the same level child wise. Its a difficult bug only to find out why proper tables header/body assignments make a difference and why is it different. In absence of finding the real cause and we just bastardize the tables because browsers are dumb enough to believe tbody suffices and what browsers add on reflow when its missing.
Ide like to test the tfoot theory, because in reality this is what it should be for buttons.
This is on a TL-WR1043Nv5 that I've had trying to get to flash DD-WRT and BS finally fixed the firmware images. Sorry, forgot to specify router model This device has no USB port, so no telling where that anomaly is. I've cleared and overridden cache, does not change anything.
Using dev tools its likely to give some indication where that is from as the <!-- --> are rendered comments inline.
Also the info page has a similar ifndef(USB so it be intersting to see if that issue is present there also.
Code looks OK above and USB button is not showing
Quote:
If you're referring to either the info.htm (Info Site) page or Status_Router.asp page, no similar rendered anomalies present. It's specific to the Diagnostics.asp (Administration -> Commands) page, and the "Save Custom" button is missing. See attached screenshot for "view page source".
We're enjoying a "Dad can't do certain things very well today" day here so I'm posting for him.
Screenshot at 2022-09-28 11-44-17.png
Description:
Screenshot of associated page source.
Filesize:
359.53 KB
Viewed:
1352 Time(s)
_________________ "The woods are lovely, dark and deep,
But I have promises to keep,
And miles to go before I sleep,
And miles to go before I sleep." - Robert Frost
"I am one of the noticeable ones - notice me" - Dale Frances McKenzie Bozzio
Joined: 31 Jul 2021 Posts: 2146 Location: All over YOUR webs
Posted: Wed Sep 28, 2022 17:15 Post subject:
<!--<!-- yep, the end also shows -->--> which is kind of why its not rendering correctly, but since code is fine, one would have to look which open/close as my initial post is invalid for devices with no usb port.
Same problem on 49830 on TL-WR901NDv2, so it's not a new bug:
administration_commands.png
Description:
Filesize:
63.02 KB
Viewed:
1282 Time(s)
administration_commands_pagesource.png
Description:
Filesize:
114.48 KB
Viewed:
1286 Time(s)
_________________ "The woods are lovely, dark and deep,
But I have promises to keep,
And miles to go before I sleep,
And miles to go before I sleep." - Robert Frost
"I am one of the noticeable ones - notice me" - Dale Frances McKenzie Bozzio
Joined: 31 Jul 2021 Posts: 2146 Location: All over YOUR webs
Posted: Wed Sep 28, 2022 20:16 Post subject:
Same issue so one of the code open/close is bogus with the usb stuffs. It wont be a new bug since this stuff has not changed in a decade or more on this file. Changes to the file are meaningless to this, so something either changed elsewhere or this has been a bug for a long long time. Wouldn't surprise me, quite a few decades old bugs of this type were only recently (this year) hammered.
We would have to back-track to the first build that fixed flashing / upgrading the TL-WR901NDv2 (r47032/3) and check for this, if that is the case. Lead Developer Mr. Gottschall has been contacted about this issue already. _________________ "The woods are lovely, dark and deep,
But I have promises to keep,
And miles to go before I sleep,
And miles to go before I sleep." - Robert Frost
"I am one of the noticeable ones - notice me" - Dale Frances McKenzie Bozzio
Joined: 05 Oct 2008 Posts: 666 Location: Helsinki, Finland / nr. Alkmaar, Netherlands
Posted: Wed Sep 28, 2022 23:49 Post subject:
the-joker wrote:
Looks like a cache issue, old code cached headbutting new code, nothing to solve.
...
I tried changing to Chromium which did not make much difference. The rightmost columns were sometimes sorted and then sometimes not. The other columns were always OK.
Then I went back to Mozilla Firefox (Linux Mint 20.3 system) and cleared the cache using its GUI. No shortcut to be found in the GUI, but apparently Ctrl+F5 is just that and is different from the reload icon on the GUI.
In Firefox it now works as it should.
'Old code headbutting new code' sounds like reload does not reload everything from the source and run(?) every bit of code to produce the page.
(I don't know much about HTML currently works as I only ever wrote very simple HTML code in the early days. Back when at my job they made secretaries edit HTML code with a text editor ... which in at least one case resulted in an irate fear of computers and keyboards, and early retirement after a long sick leave.)
I haven´t had problems with Firefox before the recent sorting issue (the bug some days ago) even though all I am used to do is reload a page.
Joined: 31 Jul 2021 Posts: 2146 Location: All over YOUR webs
Posted: Thu Sep 29, 2022 11:53 Post subject:
Reload or simple refresh or F5 (all the same action) uses cached content where possible to speed up loading any given page. It does not clear anything browser cache related.
Browsers cache CSS/HTML/JavaScript and other resources for all websites to make them faster to load, cause local content is faster than downloading new content and local always wins, problem is DD-WRT is not a regular website and is undergoing work to the HTML lately, CSS is rarely touched now, so old cached files conflict with news ones.
Your only solution is to CTRL+F5 to force clear cache, this is well documented in the build threads when there are UI issues present, this is the goto action.
I can tell you it works now no matter browser used as long as cached content is cleared.
There are tricks I could have used, but where time is limited I opted to spend my time on more productive side, asking users to clear cache seems a fair exchange of effort.
dale_gribble39 wrote:
We would have to back-track to the first build that fixed flashing / upgrading the TL-WR901NDv2 (r47032/3) and check for this, if that is the case. Lead Developer Mr. Gottschall has been contacted about this issue already.
You can and it will still be there, I can say this with utmost confidence, the acid test is any other non usb port device and try a router database build, for my money it will still be a bug there. Literally no code Ive touched directly or indirectly has any effect on this. Guaranteed as evidence by the file history. This code is decades old.
Reading the Diagnostics.asp file, and because the double rendering of comments is to blame as shown on the devtools on both screenshots provided (<!--<!-- some code here-->-->) around the fieldset (double bubble on these is invalid btw) but we need the nvm call for devices with USB, so the fieldset shows with saved parameters. ill do a patch and see what the headmaster says
Well, the solution to the cosmetic page rendering issue was to create a special conditional compile rule for src/router/httpd/visuals/ejs.c, change the <% ifndef("USB", "-->"); %> tags, and remove the <% nvm("rc_usb", "", "<!--"); %> tags from src/router/kromo/dd-wrt/Diagnostics.asp rather than creating new base.nvramconfig, etc files for nousb firmware images or other conditional compilation mechanisms. Your patch didn't remove part of the underlying problem (the presence of rc_usb= nvram parameter). <wink>
We'll check for 50300 alpha later tonight 😎🎸🎶🤘💥 _________________ "The woods are lovely, dark and deep,
But I have promises to keep,
And miles to go before I sleep,
And miles to go before I sleep." - Robert Frost
"I am one of the noticeable ones - notice me" - Dale Frances McKenzie Bozzio
Joined: 31 Jul 2021 Posts: 2146 Location: All over YOUR webs
Posted: Thu Sep 29, 2022 19:50 Post subject:
My patch would have worked fine for 99% of users, except for really edge cases where users may import configs from other routers or manually set rc_usb variables via CLI.