(old)Puppy Linux Discussion Forum

28 Mar 2014
I'm back for a while, until work takes over again.

I've built and internally tested V910 and the only thing I've found is it seems to take longer to do anything in the shutdown menu so it's now up to you to load test it. If nothing serious appears by Easter I shall move it to Release status then.

Have fun!

15 Aug 2013
Latest JWM Test Version is now attached. JWM Release 2.2 is now a month behind schedule while joe fixes the last few problems and changes he wants in it so I don't expect many more unless major stoppers appear.

21 July 2013
Latest JWM Test Version is now 835, could our non-english reading puppians please check their languages out. Thanks.

Many Languages are still wanted if anyone has time please. the pot you need to work with is in L18L's post (the fourth one down) not puppy's as we only use a subset. http://www.murga-linux.com/puppy/viewto ... 994#709994.

7 july 2013
Latest JWM Test version 827, fixes and a few new languages included.

4 july 2013
JWM Test version 819 has been uploaded as 815 ocassionally had problems with windows in the tray, sorry.

25 june 2013
JWM Test Version 815 has been uploaded.
As usual, providing no serious problems are found, I intend using this for the next release version in about seven days so can I have responses in the thread with the puppy's tested against listed please.

22 june 2013
We have a new test version, it includes fixes and additions.

15 June 2013
I have uploaded a test build of v777 for your testing. If all is ok after some user testing.

04 june 2013
I have uploaded a test build of version 768 for testing.

30 May 2013
I have uploaded a test build of jwm at version 762 for testing.

My initial reaction is that pages and windows seem to be being redrawn quicker, looking at Topview it also seems to use less cpu%age when the system is dealing with a decent load, other than those I have not found any differance yet.

15 May 2013
I have updated jwm to a release version build of 755 although I have yet to build the new x86_64 version.

It appears in two packages, as I have had requests to separate the NLS so a few people could see what languages were inside it and 'create' their own language files to be added into the set. I don't intend to carry this practice on by default as it is not normally done with a jwm build. Could those creating the language files please attach them to their responses and i'll see that they get back to joe for inclusion in the main package source.

7 May 2013
I have added below a x86_64 pet for TESTING ONLY by the fatdog64 and slackbones community. Although normally it would be best for them to add mick's jwmconfig130213.pet (page3 for url) as it has been updated to match the new jwm's config for 32bit puppies only, it's not an exact match for the way fatdog64 and slackbones64 and it does show an error on start so for now I won't recomend it be installed.

@32bit users and others, please note that mick has created a new jwmconfig2, it's at/near the bottom of page 9 and will be needed for V750+.

6 May 2013
Test Version 750 uploaded, if it's without problems next week I shall rebuild as a Release.

3 May 2013
I have uploaded a test version of JWM Version 746.

21 Apr 2013
I have rebuilt jwm's v731up to my release standard.

I have also built a new test version v733, notes are at the bottom of page7.

14 Apr 2013
A new test version 731 has been uploaded.

Please check my note at the bottom of page7 before use.

12 Apr 2013
Production Version is still 704

A new Test Version 726 has been uploaded. If no problems appear in the next week I will rebuild it at Release Standard.

This new Snapshot version has been classed by joe as a Master/Head Build and could be used by joe for his next full release (2.3) so I expect it will become the Production Version for a while.

05 Apr 2013
Production Version is still 704

A new Test Version 722 has been uploaded. If no problems appear in the next week I will rebuild it at Release Standard.

28 Mar 2013
The production Version has updated to 704.

The Test Version has updated to 716. Those new to testing please remember to use mick's jwmconfig2 mentioned and linked on page3.

There will also be a changelog on the message on page 7 for those wanting to see what has happened.

17 Mar 2013 The production version at this time is still jwm2-691.

Version 704 is at TEST ONLY status until the 24th when it will be rebuilt as production if no problems have arisen.

I have uploaded a new version JWM2-704 for final testing only, please test and report asap. It contains the 'core only'. Anyone new to testing jwm2 please also download and install micks (01micko) config pet mentioned on page3 as it's a co-requisite (if you have not already done so)!

Notes start on page5

Please make sure you include what puppy's you have tested on with your reply's.
-------------------------------
I have built a final release version of jwm2-691 and removed the previous versions.

The new version is in two pets.

The first pet, jwm2-691-w5c.pet contains only the core jwm component and docs and for someone upgrading should be enough for those upgrading from V679 or later.

The second pet, jwm2-config-691.pet contains the config components and will be needed for anyone upgrading from a version pre jwm2-679 or working from woof.

Alternately to using the second pet, you can instead use 01micko's jwmconfig2-130213.pet mentioned with a link on page 3 which works as well and has some additional functions which I do like.

---------------------------------------------------------------
I have a JWM upgrade at Version 691 attached below for you all, that seems to be ok and without that annoying window outside the screen, please test and report.

There is only one this time.

The thread picks up around the bottom of page2 for v691.

I have left version 679 in case anyone has to step back.
-------------------------------------------------------------
I have been testing for Joe Wing who has come back to working and improving jwm, in building an updated jwm for puppy. I think we have a working one now, however he is still improving it and I shall update on intervals.

For my point, it's primarily for my Mage2, as the standard one did have problems, but I will build and test against as many puppy versions as I can when I have what I consider 'puppy stable' releases.

So far there are mage2 and wary/racy versions available, the later has been tested against others and seems to work happily so for now I shall only build the two.

There are a couple of changes puppy woof builders (not users) will need to do with using the standard /etc/xdg/templates/_root_.jwmrc file which I have needed to altered slightly, as he has changed a few things. I'm not sure if the gtk people (thunor?) need to know, but I will add a note in his thread.

To quote joe

"With the latest snapshot, the number of user-configurable layers has been reduced to three ("above", "normal", and "below"). In addition, there is now a configuration option (the "border" attribute on the "TrayButton" tag) to set whether tray buttons have a border rather than the rule where tray buttons that are smaller than 48 pixels are flat".

There are other changes as well that affect the appearance slightly. However I believe to the better, but certainly the new build is so far both faster as well as more pleasing to use!

The current version I have built and tested against is 679 and is attached for others to play with. If your not using mage2 please use the jwm-679-w5c pet as the mage2 version is built specifically for mageia puppy's.

One thing you need to be aware of is that as part of the update fixes, jwm now passes application errors on rather than 'absorbing' them as it did in the past. This means that any application that has a problem and is opened under jwm will have it's errors passed onto the /tmp/xerrs.log file, this file will grow and needs ocassionally checking. I have found it interesting in that some errors have appeared here that were not found elsewhere. The problem is that some are not really relevant, being things like notices of lib links instead of the actual lib files being used, however it is finding errors and those found should be fixed, and this may only require a recomilation of the package to deal with updates, or the addition of a missing lib file that was not realized as missing![/url]

I just thought you should know, I smiled to myself when I read this.

JWM vgit-679 by Joe Wingbermuehle
compiled options: confirm fribidi icons png shape xft xinerama xpm xrender
system configuration: /etc/system.jwmrc (not the case, although the file serves as an example)
___play-tested in racy 5.3.90, kernel 3.0.25)___

"With the latest snapshot, the number of user-configurable layers has been reduced to three ("above", "normal", and "below"). In addition, there is now a configuration option (the "border" attribute on the "TrayButton" tag) to set whether tray buttons have a border rather than the rule where tray buttons that are smaller than 48 pixels are flat". Yep, simplified layering prompted a few edits for groups in my xdg template (which is customized for a jwm 'base' section of the menu, so I kind of crosschecked the given template in the dotpet rather than do a typical click-install which would overwrite my mods).

Good having a man for this, even if a few puppy-specific items like using jwmconfig for the basics could use clarification (did that in my personal man2html page). Thanks

Also of note - mouse pointer over titlebar buttons now uses vertical resize glyph on min, max and barred circle glyph on close (cursor set cw) .. m2 for move, m3 for windowmenu on the titlebar and the secondaries for titlebar min (m3 toggles shade) and max buttons work okay in normal jwm.

Some oddness with xcompmgr -n compositing .. incomplete draws of max'd windows (appears to retain size geometry in visible draw), a ctrl+s then shows window decor (in the case of xfw text editor) and after size restore remnants of the max'd window 'serial' translucency seems to follow layer rather than focus protocol (may be since I use sloppy focus or it's been that way anyhow) - serial translucency works even without Opacity tag(s) in WindowStyle

Rootmenu no longer has beveled border! However, selected menu items no longer have rounded corners and they're pretty much gone for everything except window borders - no biggie save for a consistent overall look (even xpad's decorless windows when pinned, but then xpad has good hinting anyhow)

As a final note, jwm679 seems to have a somewhat heavier footprint in memory now, roughly 3.7-1.8% (after a few xerrs.log jwm-related fixesº) in htop for me as opposed to 0.8% for jwm505.

º - yep, "jwm now passes application errors on rather than 'absorbing' them as it did in the past" and this can be a good thing, tempered by one's ability to fix various errors. Occasionally I find it to be a judgement call if things are otherwise basically functional - for instance, dillo threw up a few dpi errors and sometimes there were badwindows and so on.

thanks 'f00, slightly rounded buttons will be back with the next version I build, joe has fixed them up for those who missed them.
The current build git is only v681 as he has fixed most simple things now, he is now concentrating on the hard stuff I believe.

I shall give it a week or two/three before I look to build again.

scsijon wrote:To quote joe

"With the latest snapshot, the number of user-configurable layers has been reduced to three ("above", "normal", and "below").

I'm curious to know - do any of you guys find this really disappointing? I used to wonder if I could hack it down to 3 myself .

Joe wrote:In the near future, I intend on making some more changes including the removal of the ability to drag windows to different desktops

I will be very glad for that

scsijon wrote:certainly the new build is so far both faster

Really? What is faster about it?

switching screens requiring redraws of complicated cad images (my favorite test), opening new window apps when large numbers of windows already in use, and it also seems by observing the processor graph that the cpu load is smaller/shorter (to me anyway).

In my test system anyway with the comparison against jwm2-574 (my standard).

I wouldn't have thought that would actually have much to do with the window manager...

Hi all,

I am testing this new version in Wary 5.3

Observations:

• --menu entry icons have better clarity and show up more nicely,
  --removal of beveled edges in menu entries is consistent with Puppy design,
  --if running a custom menu, one needs to edit _root_.jwmrc and run fixmenus command,
  --application windows can still slide under the tray; apparent JWM bug,
  --memory usage is significantly greater.

On the last point, I used pmap to compare the default jwm-574 with the new jwm build where PID is the process identification number given to jwm, and I found jwm-574 used 6832K at boot-up while jwm-679 used 12980k after a re-boot... an increase of 89.9% in memory usage.

I was hoping to see a fix for the problem of windows sliding under the tray, but maybe this is part of "the hard stuff" that scsijon mentioned and will be resolved in a later build.

I am also wondering if it is possible to compile a leaner build of JWM to reduce the memory footprint.

Edit: Due to having had a "senior's moment" I fixed my math error regarding percentage increase in memory usage. However this value may not reflect a fair comparison between the two builds given the fact that both were not compiled on the same version of Wary. Further, the former build could be leaner if it was compiled with fewer options.

Monsie

Monsie wrote: --application windows can still slide under the tray; apparent JWM bug,

Why is that a bug? Windows can slide off the top or the side of the screen as well, but that isn't a bug...

disciple wrote:

Monsie wrote: --application windows can still slide under the tray; apparent JWM bug,

Why is that a bug? Windows can slide off the top or the side of the screen as well, but that isn't a bug...

Actually, I am not entirely sure that it is a bug, which is why I used the word: apparent and it's based on some written feedback by BarryK quite some time ago about outstanding issues with JWM. I may have misunderstood the point(s) he was making, but I got the impression that the desired behavior is that the application window would either meet the tray, or, slide over the top of the tray rather than under it. If I can find the particular piece that BarryK wrote, I will provide a link. In any event, I think this behavior matters most when one has several windows open and is multi-tasking, switching between windows etcetera. Ultimately, one wants to be able to quickly and easily access all of the features within a window without having to drag it into place or view because it was partly hidden.

Monsie

Monsie wrote:If I can find the particular piece that BarryK wrote, I will provide a link.

Ah, it's in a "JWM buglets" section at http://puppylinux.com/development/projects.htm
I thought you were suggesting that windows should be allowed to be moved over the tray, not under, or that neither should be allowed. But I think Barry was saying there is an issue with windows opening partly under the tray, when the screen is big enough that they don't have to. If this is (still) occurring you should check that Joe is aware of it. It sounds like it would be a reasonably simply fix... I guess that the windows are still high enough that they would be fully visible if the tray wasn't there, is that right?
I don't think this was a problem in the old days - I used to have a partial width tray at the top of the screen on autohide, and as a result windows would not be placed in the top 21 pixels of the screen, which was a little annoying

Thanks, disciple

Yes, I believe that is the link with Barry's comments... I couldn't locate the source when I was making my post. I think the issue had to do specifically with some application windows opening at the start with some of the pane under the tray. If I remember correctly Remove builtin packages used to do this, but works fine now, so maybe this issue has already been addressed.

In any event, some issues come down to user preference rather than being an outstanding bug, in which case hopefully one is able to make adjustments.

Monsie

monsie and interested others,

--application windows can still slide under the tray; apparent JWM bug,
--memory usage is significantly greater.

I have passed these two onto joe for comment as well as giving him this thread link again in case he wishes to comment directly, if not I shall pass resulting feedbacks on.

My understanding of barryk's worry was that it was already off the edge when the application window was opened, I think joe fixed that at the time.

p.s., a lot of what joe seems to be doing at the moment is code cleanups and that has to be to the good.

My understanding of barryk's worry was that it was already off the edge when the application window was opened, I think joe fixed that at the time.

scsijon,

I agree. When I mentioned this issue, I couldn't remember the exact concerns expressed by BarryK regarding JWM. I should have researched this further before posting... thankfully disciple was able to find the particular comments BarryK made to help clarify the issue. In further testing, I am not seeing any applications that open in a partially hidden state, (I already mentioned Remove builtin packages ) so I am willing to consider the issue as fixed.

The increase in memory usage by JWM is a concern... Is Joe adding a number of new features I wonder?

Monsie

When you're comparing ram usage, are the two versions you're comparing both compiled on the same system and with the same options?

As far as windows sliding under the tray is concerned, by default, trays have a higher layer than application windows. You can change the layer of the tray with the "layer" attribute:
In general windows shouldn't open under the tray unless the application requests that position for some reason or the window is too large to fit on the screen.

As far as the memory footprint is concerned, I would make sure JWM is configured the same way (as disciple noted). From my testing, newer versions actually use less memory. (Perhaps --enable-debug is in use?). There are a few new features, most significantly is SVG support, but that shouldn't affect the memory footprint much unless a lot of SVG graphics are now being used.

- Joe

Thanks, Joe

I have edited my initial post in this thread regarding memory footprint for accuracy, and as an effort to keep things in perspective.

Monsie

Thanks Joe, JWM is one of my favourite window managers. You do a great job developing it.

...and many thanks for fixing the java text focus issue.

thanks joe for a direct reply.

And, no I don't compile with debug on, you are one of the very rare application creators that I don't believe it necessary as you have created clean output each time I have compiled your code, even when I have found a bug in running the application later!

As far as memory footprint is concerned, I do compile, as 'f00 noted, with the confirm, fribidi, icons, png, shape, xft, xinerama, xpm and xrender options on. Maybe the other build mentioned didn't include some of them?

@all, Joe seems to have slowed his fixes down, if nothing new 'arrives' over the next week that may be problematic for puppy I shall build a fresh testing set next week so you should see a release build the weekend following! Also, as I was pm'd on and was my error, I will use barryk's package name codeing which means it will become jwm2-version-compiledin.pet as we are working with jwm2 not jwm1. I will build with at least three bases as I want to try barry's new Precise out.