Messages

106

Beta / Re: version 2.4

« on: August 02, 2020, 11:04:10 PM »

https://github.com/rejetto/hfs2/releases/tag/v2.4-rc07

For load testing, I asked of it several impossible things simultaneously.  It worked really well.  It did large file-list, large searches, photos, music, etc... That was just the initial test. 

Also checked with the default template and see that it was revised to archive selected.  This is good.  However, I did not get any archives.    And, then I tried again, selecting just 2 files for archive.  Didn't get an archive.   Also it didn't get stuck.   So, mostly good news. 

107

Beta / Re: 2.4 beta, freezes when there are many files

« on: August 01, 2020, 11:18:29 AM »

I can't solved it and i missed one thing, if i add other files under the real folder, problems occur. also v2.3 is fine.

To see if the difference was template, can you try HFS 2.4 RC7 + Throwback template?

108

Beta / Re: 2.4 beta, freezes when there are many files

« on: July 31, 2020, 10:45:09 PM »

Freezes when browsing if there are many files in the Real folder (about 250 files and folders). v2.3 is fine. in more detail, the thinking state continues almost infinitely and freezing or lagging.

To fix, just make a new VFS, and also clean/default the settings.

And this is a problem from previous version, if upload speed is more than about 30MB/s and there are many connections (about 10), often server is dead without error message.
but the menu interface works normally. and minimizing instead of tray icon is a little less. my ram is 16GB and internet line is 500Mbps. in Korea, 200~500Mbps is common.

For HFS, you can use a 2nd network interface, such as PCI or USB to ethernet set to 100 megabits, fix address, use just for HFS. Or use netlimiter software.  Also, to automatically get unstuck, use the WatchCat script--it has fair documentation so you can adjust the activities and timeframes to suit. 

109

HFS ~ HTTP File Server / Re: Controlling the Cache-Control header

« on: July 29, 2020, 10:19:26 PM »

Throwback for HFS 2.3 has used headers like these:
main
{.add header|Cache-Control: public, max-age=200.} and this <meta http-equiv=CACHE-CONTROL content=PUBLIC>

error messages and upload
{.add header|Cache-Control: no-cache, max-age=0.} and this <meta http-equiv=CACHE-CONTROL content=no-Cache>

See how they are equal yet opposite?  The 'mirror image' commands are necessary for the countermand/override concept to work. 
This is more common with style = old-way; new-way, and in that order; so, the browser gets as far as it can go.  The concept is actually the same.


For HFS2.4 there was slight difficulty in caching main page and then not caching login-name (some odd battle).   And, I know that those meta tags aren't right for XHTML nor HTML5 (those metas need a wee update).  Also there is no </meta> thus no /> because meta are not closable.
I got the caching to work for HFS2.4 with some experiments.  Somehow, the old style %login-link%%loggedin%%upload-link% was resistant to caching (some difference), so, it didn't cache wrong login data.  Would be nice if I had done that on purpose.  Unfortunately, I don't know how to repeat that yet.  Will investigate.


Usually there is no problem, with turning caching on--the problem is turning it off, such as for new folder (make it show) and login name (change when login) because the file-list page which should be cached (for a couple of minutes), also should change if new folder or login difference.  Very puzzling! 

I can report that caching headers work and that the problem of mixed data-to-cache + data-not-to-cache on the same page is not specific to HFS.

110

HTML & templates / Re: Something about "Dual-core browser" support (and correction)

« on: July 28, 2020, 04:42:22 AM »

Except using old-old coding standard, anyway to solve these tough problems?

You could change doctype to XHTML transitional.  And, also close a few tags for compatibility. 
After that, if anything causes breakage is was: the client's javascript is too old.  For that problem, you can put an else message:  Please update browser.

Edit (forgot to mention): 
You can do style = old way ; newer way ; newest way
In that order. 
So, if 'newest way' was rejected, then fallback can catch it. 

111

HTML & templates / Tiles and Stripes, the gallery view

« on: July 26, 2020, 11:51:14 AM »

For HFS 2.4 RC5/6/7...

Tiles-n-Stripes template is for development/testing purposes (not for main server). 
Tiles/Gallery layout may cause a couple of problems:
1).  On a fast connection, the amount of data (if page scroll+download/archive) can jam the ui and get stuck.
2).  On a slow connection, the amount of incomplete requests can snowball an overlarge 'todo' list; also stuck.

For problem#1, it was rewritten with micro-delay timers between retries and between rows of lazyload photo tiles.  This avoids the fast-connection-stuck problem, by providing some slack-time for other functions to run (ui-screen-update, list, uploads, multiple-users, etc.).   So, it can work for localhost/local-lan test and other fast connections.  Do see the updated lazyload script.

For problem#2, your clients have a variety of connection speeds, so this template might not be suitable for use over the internet. 
Possible fixes include ondemand (not implemented) whereby no tiles are shown until user clicks to see tiles on current page.  And, speed adaptive tile size, whereby the tiles get much larger size and margins for slow connections, so they could load faster if fewer fit on screen (not implemented because no client speed measure available).  If you want to put it to work, you could use the WatchCat script. 

112

HTML & templates / Re: Responsive small screen template

« on: July 26, 2020, 11:23:00 AM »

Tiles/Gallery layout is particularly difficult for HFS, with at least 2 problems: 
On a fast connection, the amount of data (if page scroll+download/archive) can jam the ui and get stuck.
On a slow connection, the amount of incomplete requests can snowball an overlarge 'todo' list; also stuck. 

Main fix (for template):  Ondemand--no tiles shown until user clicks button to enable higher load for only current page. 
-If ondemand, then you don't need to block tiles for mobile.  Sometimes, even mobile users want to see tiles (just not every page).

Is there a javascript method to Coarsely estimate client speed individually?   If that was possible, then the gallery view can be made adaptive to speed by regulating image and margin sizes (adapting to how many can load in ~2 seconds).   

113

Beta / Re: version 2.4

« on: July 24, 2020, 08:52:16 AM »

Archive-only-selected will get the job done. 
Instead of a default to archive whole server (error), just move the button (gone from top, move to selection sub-menu),
replacing archive-all, with selection-only archive instead, is enough to mainline HFS2.4.

I did not discover any other error.  But, I did discover that RC7 runs faster.  Kudos!!

114

Beta / Re: 2.4 template-making guide

« on: July 15, 2020, 11:22:11 PM »

Thanks DJ! 

115

Beta / Re: version 2.4

« on: July 15, 2020, 11:21:39 PM »

did it work on the pixel??

There are 4 templates that work for HFS2.4 (default, takeback, throwback, mobil-light), and they all work for android phone/tablet/mobile. 
The HFS2.4 new login is actually better for mobile compatibility; and, that's why the form-based login and other new work started.

116

Beta / Re: 2.4 template-making guide

« on: July 14, 2020, 08:10:43 AM »

How do I do a streamlined macro, such as:
if this and this and this all true, then do task, else do other

This could be useful for validation. 
It seems best to validate for at least 2 favorable conditions before doing upload:
%number-addresses-downloading%*%speed-out% < 7500 AND %connections% < 40
then do upload, else javascript 5 sec recheck. 

Throwback14 initial and current check only %number-addresses-downloading%*%speed-out% < 7500 (somewhat less than max workload) then do upload, else recheck in 5s.  That came from  Throwback13 gigabit support. 
Trouble is that 97 ongoing connections on slow-rate internet (workload undetected) could still stop an upload. 
So, it is necessary to test for more-than-one favorable condition before making an upload.
It is straightforward to do if this if then else, but I'm not clear how to macro for if these (several), if then else.

P.S.
Do we have %number-addresses-uploading% or other clue if there were ongoing uploads?  Because I'd like to do, if ongoing uploads, hide archive button. 

117

Beta / Re: version 2.4

« on: July 09, 2020, 06:14:08 PM »

Archive message is good, but 'Cancel' button is missing.  So, the appearance is:  only choice is click ok? 
Also probably don't need archive at root (a curious person like me, might click it just to see what happens).
Idea!
I think that you could relocate the archive button to the "Selection" sub-menu (archive can be next to the move button); so, then we can archive only what is selected.  That would be helpful. 

118

Beta / Re: version 2.4

« on: July 07, 2020, 04:55:02 PM »

a preview of the next "no jquery" default tpl.

Just 2 things:  archive is unlimited and the per-file menu has some undefined
For archive limit, I suggest, no archive at root, no archive if current folder is already 4gb, no archive if more than 1 download, no archive if more than 0 upload.  That isn't enough, but it is better.  Could add a 'manual' validation with the pop-up message stating that it could take a while to get a .tar file (most users will click 'cancel' at that point because they don't want .tar file anyway). 

119

Beta / Re: version 2.4

« on: July 07, 2020, 03:15:03 PM »

i don't understand what you mean by "validation data", and how people can benefit from removing the "speed limit" option.

The 2 speed limit and 2 connections limit cause extra work for the server and sometimes cause faults.   
So, the expectation that these settings would help, is too-different from their actual functionality (more stuck/dos).

Alternative instead of removing:  Could do just one sort of limit (if speed limit then no other limit, if connections limit then no other limit). 

120

Beta / Re: version 2.4

« on: July 07, 2020, 06:41:10 AM »

Feature redaction request:
Speed limit
Speed limit from single address
Max connections
Max connections from single address

I'm asking that these be deleted from HFS2.4's menu, because they should be a validation data, not a limiting factor.