Messages

• I know that update projects with tasks ready to report functionality was requested in the past and is already in the ToDo list. I just wanted to add that the Update project(s) right-click context menu entry could be equivelently available not just in the Tasks window, but also in the History window.
• History size could get few additional options to keep results for more than 30 days, possibly also a Forever option?
• There is an "Extra / Update all projects" menu entry. I think that such operation is not that common in projects' micromanagment Because I can relatively simply switch to the Projects window, select all available projects with left-mouseclick to top + Shift-left-mouseclick to bottom + right-mouseclick Update.
  Possibly a menu entry to "Report all finished" tasks a.k.a. "Update all projects with tasks in Ready to report state" would be much more useful?
• It is currently not possible to move the (project or task) line selection up/down with cursor keys, to resize the lines' selection with Shift+cursor keys (this all is already functional in the Messages window), or to select all lines with Ctrl+A or some equivalent (like Home + Shift+End keys, which works in Messages).
  Nevertheless, moving the "current" line is somehow supported in these window, because if the window height is smaller then the number of projects or tasks, the list is being scrolled by Up/Down cursor keys or PgUp/PgDown as if some invisible imaginary line would be moving them.
• In the Projects window, it is not possible to select an arbitrary
  list of projects by e.g. left-mouseclick to third + Shift-left-mouseclick to seventh - you can just select first-to-last. Ctrl+click is working correctly

Sorry for reporting also bugs in the wish list, I felt this belonged together.

My BT 0.29 seems to consume some 205 MB memory (Private, Pagefile). The Peak Working set is also around 220 MB, but settled down to 8 MB, thus it possibly happened just at the startup (a couple of minutes ago).

Is such (IMO) excessive memory requirement necessary for BT?

The application has 3 computers defined, but just 1 is reachable, 2 are and will out of reach.
Among the allocated memory, the process has 3 x 53.43 MB and 3 x 10.7 MB (possibly pre-allocated) blocks of memory (in the system commit), which seem to correspond to the 3 defined computers. Are they necessary to be statically pre-allocated in their whole size and in advance?

    

But I finally got the bug. [...] V 1.73 should solve the problem...

Solved. Perfect!

...and I added a lot of new stuff for BoincTasks.

Like "TThrottle 1.73: New: Integration with BoincTasks.. of local and remote hosts."? How far?

On the Options tab (or how is it actually called in English) I've tried to check "Enable BoincTasks" - what should I look for? After unchecking it, the checkboks remains disabled for some 20 seconds, it gets enabled afterwards.

After trying to start BT 0.29 for 5-6 times it merely said "Can't open some file" and terminated. Now it is finally running. Just one note (belongs to other section actually): the Graphs dlg is showing just black rounded backgrounds.

[edit]A minute later BT crashed after opening the context menu on some task. After a newstart, the graphs are displayed correctly.[/edit]

Something is leaking, maybe a handle of some kind.

Pretty exactly a Thread handle. A couple of them at once

I go hunting for them tomorrow. Shouldn't be too hard to find.... I hope.

I guess so, I hope I gave you a somehow good description.

Good luck!

The counts are actually nearly equal at any point in time, I just can not type that fast  
OK, actually it is leaking just a couple of handles per second, while throttling 5 processes, no idea how many threads in them (maybe 15-16).

The leaking pattern (in realtime, at the moment, with possibly 16 threads being enumerated and handled) is: a group of 16 handles is leaked once in a 3-5 seconds.

OK, as soon as TThrottle started it's primary job - throttling the processes, suddenly the handle count rose - TThrottle is leaking Thread handles, it is leaking tens (later to hundreds (later possibly to (tens of) thousands)) of handles pointnig to each and any thread of a process being throttled  

Example (after just a few minutes of throttling):
minirosetta_2.00_windows_intelx86.exe(27660): 31892
References: 122
Handles: 121

minirosetta_2.00_windows_intelx86.exe(27660): 32388
References: 117
Handles: 116

primegrid_llr_5.09_windows_intelx86.exe(31824): 27324
References: 111
Handles: 109

The counts are actually nearly equal at any point in time, I just can not type that fast  
OK, actually it is leaking just a couple of handles per second, while throttling 5 processes, no idea how many threads in them (maybe 15-16).



This could explain resource problems I've been occasionally experiencing in the past weeks (possibly because I do rarely reboot my computer, once in a week or two).

After restarting TThrottle, the handle counts stabilized around 420, 93, 252.



I'll check the numbers more often.

I've just noticed thet my TThrottle 1.71 is using an incredible number of handles: according to Process Manager, after a week of runtime (and consuming 3:50 hours of kernel time and 1:02 hours of user time) it uses 1 330 535 Handles, 102 GDI Handles, 261 USER Handles. (For those unaware - just 10 apps on my system use more than 1000 handles, just 4 of them between 2000 and 6000 handles.) The first number does not move, the latter two numbers oscillate around the mentioned values. I've noticed it in the Task Manager - TThrottle was consuming 10.5 MB of system's Paged Pool (maybe 15% of the whole system).

This is just an initial notice, later I'll investigate when the number of handles does go up...

After "desperately seeking my rosetta@home project for a couple of seconds long", I've noticed that the projects are alphabetically listed case-sensitively, i.e. first the projects which names begin with a capital characters, followed by projects beginning with lower case characters.

Could the sorting be done in a case-insensitive way?

The import of a list of computers could be a ice feature (BoincView could search clients on a local network and add them to the list if they were running BOINC, a common password could be optionally entered in the search dialog).

I guess that the main problem to solve is that just your BoincTasks is loosing its setup. My BT correctly keeps it over restarts.

TTh 1.70 is adding two more 1º lines above the upper 10º line, instead of just one (as it was previously). Intentionally?

http://www.efmer.eu/download/boinc/test/TThrottle_SKY.rar  And this is everything you need to build the language dll.

The .vcproj file errorneously contains non-existing RelativePath="..\TThrottle\res\eFMerCpu.ico" and RelativePath="..\TThrottle\res\idr_hand_link.ico" in addition to the valid RelativePath=".\res\eFMerCpu.ico" and RelativePath=".\res\idr_hand_link.ico".
Also, the Culture="1036" is there a couple of times (instead of Culture="1051"), meaning the Throttle_SKY.rc's Culture is sometimes set to "French (France) (0x40c)" (instead of "Slovak (0x41b)"), in addition to some French remainders in .\x64\Release folder). All this was probably not intended, but does not hurt

The graph also displays an untranslated "Please wait, not enough data yet" - you know where and why  

Bsides this, the resource dlls build just fine and play nicely with the 1.70 application.

I made a small change
[...]
is
   iMax = (int) fMax;
   iMax+=9;
   iMax/=10;
   iMax*=10;
   iMax+=1;

100 -> 109 -> 10 -> 100 -> 101

That makes just the difference at 100 
It means the step is at 101 instead of 100.

Yes, this made the change. 

I'd just recommend adding a 0.5 prior to the first rounding (or maybe all float->int roundings), like
   iMax = (int) (fMax+0.5);
but this is maybe just a question of taste.

P.S. Writing any text longer then some 12 lines is a torture, the input field is being scrolled up after each typed character    

For clarification - I've meant writing a longer text into the Forum.  Could anything be done with this? 

I don't have any problems with my Firefox browser

You are right. No problem with SeaMonkey as well. Just the iExploder