Puppy Linux Discussion Forum

[technosaurus]

[goingnuts]

technosaurus: Thanks for the mod to ashirc!
Found a vncviewer that only depends on X in old debian sources - attached the viewer part of vnc_3.3.2r2.
Also went through xchat 0.9/1.0/1.2 as I disliked 1.4... ...found that 1.2 is nice and runs well in pupngo apart from some font warnings. Attached static version of that.

[technosaurus]

Just a note on ashirc and similar apps. Bash has built-in tcp facilities (ash doesn't) that can take the place of telnet/netcat.

Just echo $msg > /dev/tcp/url/port

Where $msg would be the request containing the path of the requested page and various protocol info.

The response will come back on the same path and may contain various header info.(in other words, you can't just redirect it to a file like *.tar.bz2).

echo -e "GET /path/to/file.ext HTTP/1.1\n\n" >/dev/tcp/www.someurl.com/80
cat </dev/tcp/www.someurl.com/80
(the while read trick will fail if file starts any lines with empty space, but should work if you just need the text ... Now to experiment wtith header removal)

[goingnuts]

Using x11vnc with the Xvesa build with uclibc/tinyX11 seems to fail if Xvesa is missing XTEST extension (and MITSHM if you do not start x11vnc with -noshm switch). If Xvesa is build with -DXTEST and -DMITSHM x11vnc works ok but it seems that the Xvesa is unstable with MITSHM (freeze). So attached a Xvesa with -DXTEST...seems stable after running a couple of days with it.

[technosaurus]

maybe it wouldn't be too hard to invert the logic for x11vnc to require -shm to use shared memory instead of -noshm to disable it?

btw, have you been able to build a working Xvesa with tcp removed? I tried but even starting with -nolisten tcp failed ... it seems like maybe something is not ifdef'd. There were a couple of other things that may be easier to do than this if I could just find them - like starting in the highest supported resolution and bpp instead of a fixed one (the default doesn't work well on _all_ systems, but using the highest supported one _should_ ... in theory)

[technosaurus]

our alternative to gtkdialog has been right under our noses (tcl/tk=8.5.X)
tk apps use about the same RAM as rxvt and there are a bunch

tcl

[goingnuts]

[technosaurus]

re: tcp - building without -DTCPCONN failed ... I'll try it again though, I may have removed other defines too.

tcl/tk really should just combine altogether they are only a single test away from being able to be a multicall binary
if you do a diff of tclappinit.c and tkappinit.c ... essentially tkappinit.c is tclappinit.c with an extra code block to run the tk interpreter

continuing on further the hwish in hv3 is tkappinit.c with an extra line to run the tkhtml interpreter
(I haven't explored this any further, it may be this way for other compiled tk apps)

... so we could determine which interpreter(s) to run based on basename(argv[0]) as in the other mcbs

[goingnuts]

Turns out that msgcat is an internal tcl-function...
Also seems that if you have a static build of tk tclsh is not needed.
So attached a stripped down pet (removed man/encodings/msg etc) with tcl/tk 8.5 static build (uclibc/tinyX11) that runs all demos of tk witout errors in pupngo (no shared libs at all needed). The demos and images-folders can be deleted - saving approx. 900K on disk - but included to have something to test with - leaving us with a total size on disk for tcl/tk approx. 1700K (if wish is upx ultrabrute compressed...).
Now we only need to find all those tk apps to replace the gtk1 apps (editor, ftp, archive, mail, paint, chat etc.).

20130513: Removed attachment as forum does not support file size anymore.

[PANZERKOPF]

[sc0ttman]

[technosaurus]

This is why I never get anything accomplished...

I was in the process of learning tcl/tk and after pouring over several examples, realized a major portion of its power lies in the ability to load modules on the fly (it is still excellent even without dynamic loading).
needed for tcl-gtk-0.5, hv3, filerunner, brlcad and many other top tk apps

This made me rethink the mcb direction - what difference would it make to uses shared versions of the tiny libs?
... my gut check says it could be an overall win, but I would probably end up wanting to have an xz compress squashfs image icluded in the kernel's initramfs so that the entire thing does not get uncompressed with the kernel, but then I would also need bootstrap code to mount the squashfs image on root

so, knowing I would need code that included "squashfs" and "#include <sys/mount.h>" ... I googled it to see what it turned up ... lo and behold I got something that even included unionfs - awesome

http://openpli.git.sourceforge.net/git/gitweb.cgi?p=openpli/openembedded;a=blob;f=recipes/dreambox/dreambox-boottool/boottool-dm7025.c

[Aitch]

[goingnuts]

PANZERKOPF & sc0ttman: Nice refs - supernotepad-2011 works but the ones needing dyn load of ext libs does not.

[PANZERKOPF]