site  contact  subhomenews

Legal declarations for EasyOS

April 14, 2019 — BarryK

Puppy Forum member 'scsijon' informed me of a project named "Easy Linux" registered on It had been registered many years ago, then nothing done, so a non-starter dead project. I contacted sourceforge and they kindly removed it.

However, it got me thinking, EasyOS does need a legal statement page up-front. There is one in the welcome-page at first bootup of Easy, and scattered around in various web pages.

I have put together what I hope covers my arse ("ass" if you are American) in all situations:

...maybe I should add something about this being a totally non-commercial project? 

Tags: easy

EasyOS Forum closed

April 12, 2019 — BarryK

The time has come for another rethink of where the effort is getting directed. I closed down the Quirky project as the effort was becoming too dispersed. The same thing is happening with the port of OpenEmbedded perhaps.

And then there is the forum. I haven't been reading it. If I can't do it justice to be regularly reading it and responding to posts, then it should not continue. So it hasn't, this is now read-only:

This thread in the Puppy Forum continues:

Regarding the OE port, yeah it is nice to have control of the entire cycle, from compiling source packages to deploying a final distro, however, a lot of work went into the Thud upgrade, and I am not happy with the result.

So, the OE port might get sidelined. Continue with the "Pyro" series of the OE port, probably for another couple of years. But the next step might be to return to building EASYOS from Slackware binary packages.

Slackware is one of the few distros that meets my criteria, such as no systemd, minimal messing around with the original package (directory layout, package splitting, config files, etc.) -- basically, Slackware is plain-vanilla-Linux, conservative in every respect.
Unfortunately, Patrick, the main guy developing Slackware, has had his own share of issues the last couple of years, and the next release of Slackware is slipping away.

Talking about slipping away, the SeaMonkey developers are having problems also. The Firefox developers are obviously on a trip, introducing new requirements for compiling, such as Rust and LLVM, and other architectural changes. There were even rumblings about forking SM, but that won't happen, as their developer numbers can't maintain such a fork, most likely anyway.

For me, it is going to become increasingly difficult to compile Firefox, and SM if it goes down the same route.

I don't want to get started on Rust... they claim that it is a systems development language. Ha ha, what a laugh. There are those of us who know what a systems development language really is, it is called C, or even ASM.
What these Rust-lovers really love, is the conveniences that rust brings, such as automatic cleaning up of open streams, and handling of concurrency, things that you have to do manually in C.
For a bit of coding convenience, they are embracing a monster. 

Tags: easy

Thud goes thud again

April 09, 2019 — BarryK

Easy is using a simple HTML viewer for local help files, named "surfer", which uses libhtml, an old gtk+-2 library. However, in the Thud build it crashed. After a couple of hours puzzling, I found that HTML files with a line-feed after the "</html>" tag, caused the crash. Remove that LF from end of file, surfer works.

However, surfer in Pyro build of Easy is quite OK with that LF at end of file.

I suspect a change in gtk+-2. Pyro has 2.4.31, Thud has 2.4.32.

Then there is Xine UI, it hangs, hangs the entire desktop. Very reluctant to be forced to change media player because of that. 

Tags: easy

Can now print from Thud build

April 07, 2019 — BarryK

I posted yesterday about unable to print from the "Thud" build of Easy:

I didn't want to give up. There had to be something different from the Pyro builds, most likely package versions is the culprit.

I did a complete recompile in OE, with cups, cups-filters and ghostscript rolled back to same versions as Pyro, no joy. There was still gutensprint, poppler, poppler-data and qpdf not rolled back. Rule out gutenprint, as the PPD file for my Brother HL-2040 laser printer uses foomatic, not gutenprint.

I don't want to roll-back poppler, as LibreOffice needs the later version. Or rather, got a compile error when used poppler 0.70.0. Now using 0.75.0. Note, Pyro build has 0.53.0.

qpdf is 8.2.1 in the Pyro build, 8.4.0 in the Thud build. A quick experiment, installed the 8.2.1 binary package from Pyro, suddenly printing works!

Sigh, this kind of thing happens too often. Last year, there was a version of cups-filters that didn't work, and after reading reports on the Internet, found the fix was to roll back to an earlier version. Much later on, advanced to the latest version, 1.21.5 for Pyro, and that worked. Using that same version in the Thud build.

Just to make sure all libraries are linked properly, doing another overnight build in OE. 

Tags: easy

Fun and games compiling LibreOffice

April 07, 2019 — BarryK

Posting some notes here, in case they are of help to anyone else. It was just a couple of years ago that I succeeded in compiling LibreOffice -- prior to that, tried a few times, gave up.

Then there was OpenEmbedded, a cross-compile environment, succeeded there, as reported here:

...not without some difficulty. The version of LO in OE is 5.0.x, rather old. So, in EasyOS I have been compiling LO in a running Easy, more recent versions.

There is something very peculiar with all of the LO versions, compiling of icon-themes is broken, though other distributions do not seem to have the problem. My tentative conclusion is that Easy/Quirky/Puppy is has a Busybox applet that LO uses, but the output of the applet is not quite as expected. The same situation in OE -- it makes use of many utilities in the host system.

I recently compiled LO, the latest stable release, in EasyOS 2.0pre (Thud), and here are some notes about the hoops that I had to jump through...


# tar -xf libreoffice- --no-overwrite-dir
# cd libreoffice-

Fix for boost >= 1.69, if using system boost (which I am not, as compile failed):

# sed -i 's/loaded;/bool{loaded};/' sfx2/source/appl/shutdownicon.cxx

# mkdir external/tarballs

...copy libreoffice-translation tarball here.

# export LO_PREFIX=/usr
# sed -e "/gzip -f/d" -e "s|.1.gz|.1|g" -i bin/distro-install-desktop-integration
# sed -e "/distro-install-file-lists/d" -i
# PTNit1='s%with_theme="[^"]*%with_theme="tango%'
# sed -i -e "$PTNit1" configure

deleted all folders in 'icon-themes' except 'tango' and 'colibre'.

# mkdir icon-themes/export

# ./configure --disable-firebird-sdbc --without-fonts --without-krb5 --build=x86_64-oe-linux --prefix=/usr --exec_prefix=/usr --bindir=/usr/bin --sbindir=/usr/sbin --libexecdir=/usr/libexec --datadir=/usr/share --sysconfdir=/etc --sharedstatedir=/com --localstatedir=/var --libdir=/usr/lib --includedir=/usr/include --oldincludedir=/usr/include --infodir=/usr/share/info --mandir=/usr/share/man --without-doxygen --enable-release-build --with-parallelism=4 --without-java --disable-coinmp --enable-python=system --with-tls=nss --without-galleries --disable-gstreamer-1-0 --disable-avahi --enable-gtk --disable-gtk3 --disable-odk --disable-postgresql-sdbc --with-system-libs --without-system-clucene --without-system-xmlsec --without-system-libepubgen --without-system-libqxp --without-system-libzmf --without-system-libstaroffice --without-system-cppunit --without-system-mdds --without-system-odbc --without-system-orcus --without-system-lpsolve --without-system-gpgmepp --with-system-libodfgen --with-system-libwps --with-system-libcdr --with-system-libmwaw --with-system-libetonyek --with-system-libfreehand --with-system-libebook --with-system-libabw --without-system-libnumbertext --with-system-poppler --without-system-boost --with-lang=ALL --without-system-redland --enable-release-build=yes --disable-bundle-mariadb

Note, --with-system-boost compile fails.

edit Makefile, "check-if-root" function, change "exit 1;" to "exit 0;"

# make build-nocheck


# touch workdir/CustomTarget/postprocess/images/export_links.txt.tmp
# make build-nocheck
# touch workdir/CustomTarget/postprocess/images/tango_links.txt.tmp
# make build-nocheck

...success, 'make build-nocheck' finished.

# make DESTDIR=/mnt/sdc1/downloads2/input527/0-libreoffice/lo distro-pack-install

but, do have 'instdir' which has everything in it, so can do manual install. Note, with earlier versions of LO, the "make distro-pack-install" step did work.

Tags: easy

Thud goes thud

April 06, 2019 — BarryK

As reported in the last couple of blog posts, I have recompiled everything based on the "Thud" release of OpenEmbedded/Yocto, as the "Pyro" release was May 2017, and I thought Easy needed to be brought up to date with the latest libraries.

This has been intense work over the last three weeks. Some hurdles, but got a working EasyOS build. Then, last night decided to test printing ...broken. There is a error log file, but I have not been able to figure out the cause of the failure. Very frustrating. CUPS debugging is a "dark art", if something goes wrong, the error log needs a CUPS developer to understand.

So for now, staying with Pyro.

In the 1.0.14 release, /usr/local/pup_event/pup_event_frontend_d was not quite working right. It was written in BaCon, and I have rewritten it in C, and it now works nicely. Not putting down BaCon, but in C I am able to see exactly what is going on.

The latest PET with the pup_event executables:

The WoofE project, has pup_event source (95MB): 

Tags: easy

EasyOS Thud build progressing slowly

April 03, 2019 — BarryK

A short progress report. I am still here, working on the Thud build, which will be version 2.0.

Everything got compiled in my port of OE, and some big packages were compiled manually in a running Easy (such as SeaMonkey, Scribus, LibreOffice and Inkscape). Then I decided to bump the versions of 'icu' and 'poppler', so yesterday reran OE. Fixing some things in Woof, next have to recompile SeaMonkey, etc, due to the changed libs -- and that takes awhile, LibreOffice is an overnight compile.

I am rewriting 'pup_event_front_d', converted from BaCon to C. It wasn't working right, and with C code I can see more precisely what is going on.

Lots more things to do, so cannot really say when to expect 2.0, maybe about 3 weeks away. 

Tags: easy