|
View:
New views
9 Messages
—
Rating Filter:
Alert me
|
|
|
|
|
|
|
|
|
Re: XNEE problems / Questionshi again
Burke, James wrote: > snip > I assume you record a Gnumeric session? > > JB - Actually, gnumeric was not available. I used oocalc which is an OpenOffice application. > JB - I also tried xcalc. Both failed with the synchronization failure message above. > JB - Might the events to record vary based on the application? I tried recording all events > JB - but the amount of data became too large. > > Read this? > > http://itupw056.itu.chalmers.se/xnee/doc/xnee.html#SEC106 > > JB - Yes, I did read it. This section came across to me as more of a high level description > JB - not so much how to synchronize or what the cause might be of the sync error I received. OK. You probably record the "wrong" events. Erh, how did that sound? Since X11 is asynchronous (no way to make sure that events gets send in the right order when replaying as when recording) we can rather easy get into a situation where we get a bunch of X events in advance and miss a couple of Y events (they will come later) when replaying. Xnee manages this by thresholds, but my experience is that setting them too high will lead to "more or less unsynced" behavior of your app. So I would go for investigating what events gets sent when you test. But let's get down to your problem. I tested oocalc a bit and noticed two things: 1) Xnee seems to be doing sth wrong when reading settings from command line and from recorded file :( Q&D solution: remove the three lines in your recorded session file that sets the thresholds 2) Try synching with event 15 (VisibilityNotify) Record: ------- ./cnee/src/cnee --record --mouse --keyboard \ --delivered-event-range 15 -o hesa.xns -sk p Note: recording stops when you press the key p Replay ------- replay with thresholds set high (and one TOO high): ./cnee/src/cnee --replay -f hesa.xns --verbose -bv \ -mat 100 -mit -100 -tot 400 Note: -bv comes from a Xnee built with buffer verbose supprt. use it at will, but you need to enable "buffer verbose" during configure NOTE: I am looking deep into the sync code at the moment!!!! /hesa _______________________________________________ Bug-xnee mailing list Bug-xnee@... http://lists.gnu.org/mailman/listinfo/bug-xnee |
|
|
Re: XNEE problems / QuestionsBurke, James wrote:
> >> Also, I need the script cability to fake user input and time how long > > it takes specific tasks to complete. > > This depends on what you define as a task. A window update? A line > written in a canvas? A disk write? A program executed and exited? > > JB - Tasks of interest would be things that a typical consumer > JB - might do on a netbook or smartphone. For example, open an editor, > JB - open a file, input text, save the file, searching the file > JB - for all instances of a word, browsing a file at a certain rate, etc. > JB - I'm looking for a way to automate this as well as be able to > JB - measure performance for each of these tasks. Knowing when a > JB - program is executed, window up and ready for input would also > JB - be required, as well as determing when the program is exited. > > > Are you aware of any other tools like xnee that would > > meet my requirements? > > Depends on the above > >> Any help or insights would be greatly appreciated. Easiest would be to record/replay some "test cases" and measure the time it takes to exec cnee. This can typically be done inside a shell script, e g (bash syntax): START_TIME=$(date +%s) cnee --replay ........ lots of args STOP_TIME=$(date +%s) EXEC_TIME=$(( $STOP_TIME - $START_TIME )) _______________________________________________ Bug-xnee mailing list Bug-xnee@... http://lists.gnu.org/mailman/listinfo/bug-xnee |
|
|
Re: XNEE problems / QuestionsBurke, James wrote:
>> Shell problem: >> I followed the example in section "5.8 Shell scripts using Xnee" of the xnee manual. Part of that example included xnee_fake_key calls: >> Xnee_fake_key l >> Xnee_fake_key s >> Xnee_fake_key XK_Return > > Try "Return" (and not XK_Return) ...... > > Does it work, then one of us has made a woops in the docs. > > JB - No, Return did not fix the problem. Strange, works here. >> When I run the script, only the "l" gets outputted - the "s" and the return do not. > >> I found that if I insert a "xnee_init_file" right before each "xnee_fake_key" call, it works. > > The example and the bash file you're sourcing states you must call the > init function. > > JB - I do call "xnee_init_file" at the start of the script: > JB - #!/bin/bash > JB - . ~/jpburke/xnee_code/xnee-3.02/share/xnee.sh > JB - > JB - xnee_init_file > JB - sleep 1 > JB - xnee_fake_key l > JB - xnee_fake_key s > JB - xnee_fake_key Return > JB - sleep 1 > JB - xnee_close_down > JB - > JB - As executed above, I only get the "l" output. If I insert an xnee_init_file > JB - call before each xnee_fake_key call then it works as intended - "ls" and > JB - return is typed in my xterm and a listing of files in the current directory > JB - is displayed. odd ..... yet crappy Will have a look ... but after the other bug is fixed. _______________________________________________ Bug-xnee mailing list Bug-xnee@... http://lists.gnu.org/mailman/listinfo/bug-xnee |
|
|
Re: Re: XNEE problems / QuestionsHenrik Sandklef wrote:
> Burke, James wrote: >> JB - I do call "xnee_init_file" at the start of the script: >> JB - #!/bin/bash >> JB - . ~/jpburke/xnee_code/xnee-3.02/share/xnee.sh >> JB - >> JB - xnee_init_file >> JB - sleep 1 >> JB - xnee_fake_key l >> JB - xnee_fake_key s >> JB - xnee_fake_key Return >> JB - sleep 1 >> JB - xnee_close_down >> JB - JB - As executed above, I only get the "l" output. If I insert an >> xnee_init_file I can now reproduce the error. Will look more into it. _______________________________________________ Bug-xnee mailing list Bug-xnee@... http://lists.gnu.org/mailman/listinfo/bug-xnee |
|
|
Re: Re: XNEE problems / QuestionsThis is now solved in CVS head.
A script now can look like this: #!/bin/bash . /opt/xnee/latest/share/xnee/xnee.sh xnee_fake_key l xnee_fake_key s xnee_fake_key Return xnee_fake_key p xnee_fake_key s xnee_fake_key XK_space xnee_fake_key a xnee_fake_key u xnee_fake_key x xnee_fake_key w xnee_fake_key Return Henrik Sandklef wrote: > Burke, James wrote: > >>> Shell problem: >>> I followed the example in section "5.8 Shell scripts using Xnee" of >>> the xnee manual. Part of that example included xnee_fake_key calls: >>> Xnee_fake_key l >>> Xnee_fake_key s >>> Xnee_fake_key XK_Return >> >> Try "Return" (and not XK_Return) ...... >> >> Does it work, then one of us has made a woops in the docs. >> >> JB - No, Return did not fix the problem. > > > Strange, works here. > >>> When I run the script, only the "l" gets outputted - the "s" and the >>> return do not. >> >>> I found that if I insert a "xnee_init_file" right before each >>> "xnee_fake_key" call, it works. >> >> The example and the bash file you're sourcing states you must call the >> init function. >> >> JB - I do call "xnee_init_file" at the start of the script: >> JB - #!/bin/bash >> JB - . ~/jpburke/xnee_code/xnee-3.02/share/xnee.sh >> JB - >> JB - xnee_init_file >> JB - sleep 1 >> JB - xnee_fake_key l >> JB - xnee_fake_key s >> JB - xnee_fake_key Return >> JB - sleep 1 >> JB - xnee_close_down >> JB - JB - As executed above, I only get the "l" output. If I insert an >> xnee_init_file >> JB - call before each xnee_fake_key call then it works as intended - >> "ls" and >> JB - return is typed in my xterm and a listing of files in the current >> directory >> JB - is displayed. > > > odd ..... yet crappy > > > Will have a look ... but after the other bug is fixed. > > > > _______________________________________________ > Bug-xnee mailing list > Bug-xnee@... > http://lists.gnu.org/mailman/listinfo/bug-xnee > _______________________________________________ Bug-xnee mailing list Bug-xnee@... http://lists.gnu.org/mailman/listinfo/bug-xnee |
|
|
Re: Re: XNEE problems / QuestionsYes, finally the bug was found and fixed it.
Sync now works again. An error in the parser was the reason..... :( Will commit to CVS and make a 3.03 release. Finally /hesa Henrik Sandklef wrote: > hi again > > Burke, James wrote: > >> snip > > >> I assume you record a Gnumeric session? >> >> JB - Actually, gnumeric was not available. I used oocalc which is an >> OpenOffice application. >> JB - I also tried xcalc. Both failed with the synchronization failure >> message above. >> JB - Might the events to record vary based on the application? I tried >> recording all events JB - but the amount of data became too large. >> >> Read this? >> >> http://itupw056.itu.chalmers.se/xnee/doc/xnee.html#SEC106 >> >> JB - Yes, I did read it. This section came across to me as more of a >> high level description >> JB - not so much how to synchronize or what the cause might be of the >> sync error I received. > > > OK. > > You probably record the "wrong" events. Erh, how did that sound? > > Since X11 is asynchronous (no way to make sure that events gets send in > the right order when replaying as when recording) we can rather easy get > into a situation where we get a bunch of X events in advance and miss a > couple of Y events (they will come later) when replaying. Xnee manages > this by thresholds, but my experience is that setting them too high will > lead to "more or less unsynced" behavior of your app. So I would go for > investigating what events gets sent when you test. > > But let's get down to your problem. > > I tested oocalc a bit and noticed two things: > > 1) Xnee seems to be doing sth wrong when reading settings from command > line and from recorded file :( > > Q&D solution: remove the three lines in your recorded session file > that sets the thresholds > > > 2) Try synching with event 15 (VisibilityNotify) > > Record: > ------- > > ./cnee/src/cnee --record --mouse --keyboard \ > --delivered-event-range 15 -o hesa.xns -sk p > > Note: recording stops when you press the key p > > > Replay > ------- > > replay with thresholds set high (and one TOO high): > > ./cnee/src/cnee --replay -f hesa.xns --verbose -bv \ > -mat 100 -mit -100 -tot 400 > > Note: -bv comes from a Xnee built with buffer verbose supprt. > use it at will, but you need to enable > "buffer verbose" during configure > > > NOTE: I am looking deep into the sync code at the moment!!!! > > > > /hesa > > > _______________________________________________ > Bug-xnee mailing list > Bug-xnee@... > http://lists.gnu.org/mailman/listinfo/bug-xnee > _______________________________________________ Bug-xnee mailing list Bug-xnee@... http://lists.gnu.org/mailman/listinfo/bug-xnee |
|
|
Re: Re: XNEE problems / Questionshttp://www.sandklef.com/hesa/index.php/2009/04/19/gnu-xnee-sync-error-fixed-30296-is-out/
Henrik Sandklef wrote: > Yes, finally the bug was found and fixed it. > > Sync now works again. An error in the parser was the reason..... :( > > Will commit to CVS and make a 3.03 release. > > Finally > > /hesa > > Henrik Sandklef wrote: >> hi again >> >> Burke, James wrote: >> >>> snip >> >> >>> I assume you record a Gnumeric session? >>> >>> JB - Actually, gnumeric was not available. I used oocalc which is an >>> OpenOffice application. >>> JB - I also tried xcalc. Both failed with the synchronization failure >>> message above. >>> JB - Might the events to record vary based on the application? I >>> tried recording all events JB - but the amount of data became too large. >>> >>> Read this? >>> >>> http://itupw056.itu.chalmers.se/xnee/doc/xnee.html#SEC106 >>> >>> JB - Yes, I did read it. This section came across to me as more of a >>> high level description >>> JB - not so much how to synchronize or what the cause might be of the >>> sync error I received. >> >> >> OK. >> >> You probably record the "wrong" events. Erh, how did that sound? >> >> Since X11 is asynchronous (no way to make sure that events gets send >> in the right order when replaying as when recording) we can rather >> easy get into a situation where we get a bunch of X events in advance >> and miss a couple of Y events (they will come later) when replaying. >> Xnee manages this by thresholds, but my experience is that setting >> them too high will lead to "more or less unsynced" behavior of your >> app. So I would go for investigating what events gets sent when you test. >> >> But let's get down to your problem. >> >> I tested oocalc a bit and noticed two things: >> >> 1) Xnee seems to be doing sth wrong when reading settings from command >> line and from recorded file :( >> >> Q&D solution: remove the three lines in your recorded session file >> that sets the thresholds >> >> >> 2) Try synching with event 15 (VisibilityNotify) >> >> Record: >> ------- >> >> ./cnee/src/cnee --record --mouse --keyboard \ >> --delivered-event-range 15 -o hesa.xns -sk p >> >> Note: recording stops when you press the key p >> >> >> Replay >> ------- >> >> replay with thresholds set high (and one TOO high): >> >> ./cnee/src/cnee --replay -f hesa.xns --verbose -bv \ >> -mat 100 -mit -100 -tot 400 >> >> Note: -bv comes from a Xnee built with buffer verbose supprt. >> use it at will, but you need to enable >> "buffer verbose" during configure >> >> >> NOTE: I am looking deep into the sync code at the moment!!!! >> >> >> >> /hesa >> >> >> _______________________________________________ >> Bug-xnee mailing list >> Bug-xnee@... >> http://lists.gnu.org/mailman/listinfo/bug-xnee >> > > > > _______________________________________________ > Bug-xnee mailing list > Bug-xnee@... > http://lists.gnu.org/mailman/listinfo/bug-xnee > _______________________________________________ Bug-xnee mailing list Bug-xnee@... http://lists.gnu.org/mailman/listinfo/bug-xnee |
| Free embeddable forum powered by Nabble | Forum Help |