Puppy Linux Discussion Forum

anikin · Joined: 10 May 2012 Posts: 370

Here's how it looks in upup precise382. I used the native upup version, which is 574. Perhaps, I should ask scsijon what's the latest usable version without those pointer glyphs and do a recompile. I forgot to run "strip", so it's a bit chubby, but very fast indeed. These are the removed options:

anikin · Joined: 10 May 2012 Posts: 370

pemasu · Joined: 08 Jul 2009 Posts: 5456 Location: Finland

anikin · Joined: 10 May 2012 Posts: 370

pemasu · Joined: 08 Jul 2009 Posts: 5456 Location: Finland

Stripping makes binaries and libraries smaller. Most distros strip binaries and libraries. At least those which I have tested. For example Ubuntu and Debian .deb packages come with stripped stuff.
Puppy philosophy is to offer small size with as full usability as possible. Stripping absolutely belongs to that philosophy.

All woof built Puppies and puplets come with stripped binaries and libraries. Woof 3builddistro script takes care of that:

disciple · Posted: Fri 15 Mar 2013, 19:49 Post_subject:

anikin · Joined: 10 May 2012 Posts: 370

Here's some more:
http://software.intel.com/en-us/blogs/2013/01/17/x86-gcc-code-size-optimizations

http://software.intel.com/en-us/blogs/2012/09/26/gcc-x86-performance-hints

pemasu · Joined: 08 Jul 2009 Posts: 5456 Location: Finland

Hmm...This derails the jwm updates thread. I will post this woof code here and then stop. That is how about all woof built Puppies has been stripped.

anikin · Joined: 10 May 2012 Posts: 370

scsijon,
It was not my intention to sidetrack this thread, my apologies. I'll put it back into JWM realm with my last question to pemasu:

OK, the compilation has been completed, with all the default/predefined strip settings. I shouldn't worry, my new JWM is at its optimum, small and fast. Stop there and be happy. But I go further and run "strip jwm" and the result is a *significant* reduction in size. What's going on? What else has been shaved off? Do I still have a fully functional binary?

pemasu · Joined: 08 Jul 2009 Posts: 5456 Location: Finland

Well. Compilation does not strip anything. You need to have some code to run strip command. I mean executing make does not offer any means for stripping yet.

Jwm source code has it, when you execute make install with or without other options.

new2dir has it inside code and executing new2dir make install surely has it, in fact twice since make install execute it also.

or you do it from commandline by executing strip with or without other stripping options.

So...because you didnt tell what you actually did, I cant tell you what step you missed and you got smaller binary after executing strip command.

And yes....you get working binary after executing strip jwm

anikin · Joined: 10 May 2012 Posts: 370

My bad, I'm sorry. I haven't yet tried doing it as scsijon recommends, there was no "new2dir" in my scenario. I will redo the whole thing.

scsijon · Posted: Sat 16 Mar 2013, 07:02 Post_subject:

thank you pemasu for dealing with the strip queery.

My own take is to NEVER use a straight strip,

If part of the compile/build/install hasn't a strip step or I'm not building for puppy, where I always use the new2dir in my build, I ONLY EVER use "strip --strip-debug" or "strip --strip-unneeded", nothing else.

And NEVER USE "--strip-all" as it can make the file you have built useless for use on any machine but the one it was built on.

I learnt many many years ago that it's better to not strip than just "strip" alone. I lost a weeks build-modify-compile time for a team I was managing by forgetting to add the -- modifier. I had to restart the final build again and that took some ten hours. I was somewhat late home that night!

And lets leave that there unless any of the guru's have a further comment we should know about.

scsijon · Posted: Sun 17 Mar 2013, 01:39 Post_subject: Test Version jwm2-704 available

I have updated the test version to 704. Details and pet are as usual in the first screen on page1.

It includes the middle and right mouse button functions requested for the maximize function as well as some fixes.

Can we have as many testers as possible across all versions of puppy please as everything left on the problem system are enhancement requests.

The only thing you need to be made aware of is that if you kill rox-filer using the tray app-window menu kill you will also loose your background and it's icons, similar to the black screen of death, but your tray and apps via menu still work. To get your background and icons back, you need to restart the x-server on the shutdown menu. I did think it was a further fault, but it's not, however I have asked joe to consider if it is possible that some sort of trap function where only the rox-filer app window is killed and not the rox-filer app itself.

pemasu · Joined: 08 Jul 2009 Posts: 5456 Location: Finland

What I tested 704, it worked in my normal usage ok. The only disturbing feature is those changed mouse forms when hover over close, minimize and maximize buttons. I also felt that rectangular mouse pointer somewhat odd.

I hope there could be possibilty to change that feature back to usual non-changing behavior. This has been posted earlier so I suppose it is known what I mean.

If there is already way to change this, I would be grateful for the howto.

`f00 · Posted: Sun 17 Mar 2013, 22:11 Post_subject: Sub_title: ..

Some earlier ideas about titlebar button masks, a sub-theme to be sure.

Compiled jwm-704 in racy 5.4.91 for my own personal trial (worked fine with maybe a puzzle on "Desktop" tag use, see attached pix), same old minor issues with xcompmgr -n (may try kcompmgr just to see if there's any change there). Will also try the jwm2-704-w5c dotpet (at least the binary) as a cross-check..

Meh, the pointer-change-on-hover over the titlebar buttons is just 'new' to jwm (it's been around for awhile in some other wms like afterstep and e16 to name a couple). At least jwm's theming is still relatively simple Smile