tinycore_1.0rc9

[roberts]

Release Candidate 9 is now posted:

* Removed remnant files /etc/init.d/functions.lua and /root/.luafltkrc
* Fixed checkfs boot option to call with busybox options
* Updated boot time f3 text to specify file systems checked.
* Stripped debug symbols from depmod
* Dropped gradient from jwm to better support older hardware.
* Adjusted jwm taskbar to be consistent with the selected window color.
* Improved tce-load to disallow duplicates.
* Added 3g-modem modules
* Increased Aterm scroll buffer to 1000
* Rstored animation to wbar as the default
* Improved tce-load to store  and later process local .dep files, typical use is "Download Only" into /optional Then upon reboot, load local appbrowser->File->Install Optional select only the major apps, the dependent files if available will also be loaded. This change allows the user to keep the super fast boot times and load locally stored extensions upon demand with dependices resolved.

Files that have changed that may in your backup or persistent storage device:
.jwmrc
.jwmrc-theme
.Xdefaults

[Juanito]

up until now, I've been running a number of scripts to load various extensions after boot.

With rc9 I now get errors of the form:

Code: [Select]

...
can't open /path-to-file/extension.tce.dep
If I understand correctly, I now need to have all the dep files present, but what if an extension doesn't have any deps?

Note that if I load the extensions one-by-one from a terminal, I get:

Code: [Select]

$ tce-load /path-to-file/wireless-2.6.26.tcem
/path-to-file/wireless-2.6.26.tcem..but the extension isn't loaded - same with the app-browser loading local?

Did the gtk theme disappear, the windows look "flat" again?

Edit: Just went back to rc8 to double-check - everything works fine there.

[curaga]

The dep file's existence should be checked before trying to open it:

--- tce-load    2008-12-31 11:05:37.000000000 +0200
+++ tce-load.new        2008-12-31 11:13:19.000000000 +0200
@@ -143,12 +143,14 @@
 if [ -z "$2" ]; then
    FROMWHERE=`dirname "$APP"`
    echo "$FROMWHERE"
-   for F in `cat "$APP".dep`; do
-      DEPFILE="$FROMWHERE"/"$F"
-      if [ ! -f /usr/local/tce.installed/"$DEPFILE" ]; then
-         install "$DEPFILE"
-      fi   
-   done
+   if [ -f "$APP".dep ]; then
+          for F in `cat "$APP".dep`; do
+             DEPFILE="$FROMWHERE"/"$F"
+             if [ ! -f /usr/local/tce.installed/"$DEPFILE" ]; then
+                install "$DEPFILE"
+             fi
+          done
+   fi
 fi
 
 install "$APP" "$2" "$3"

..That does not explain though why your extensions did not get installed. Please add echo statements to every part to see where it gets.

[roberts]

.dep are not required but moving forward will be stored and used when found for on-demand local loading.
Your extension loaded but the resulting exit code to appbrowser indicated an error.
A simple if statement was/is needed.  rc9a to be posted shortly.

[roberts]

rc9a now posted.

[Juanito]

that solved part of the issue - it looks like the other part was my own dumb fault..

I was loading an extension from a folder with a dep file, but without the extensions quoted in the dep file - i.e.

Code: [Select]

$  tce-load /mnt/sda1/source/buildtc/08_10_25/wireless_tools.tcel
/mnt/sda1/source/buildtc/08_10_25/wireless_tools.tcel
tar: can't open '/mnt/sda1/source/buildtc/08_10_25/wireless-2.6.26.tcem': No such file or directory
Note that it seems that once there has been an error like the above, then "tce-load /path-to-file/wireless-2.6.26.tcem" does not work. On moving wireless_tools.tcel to a folder without the dep file and after re-boot, things again work fine.

On a different subject, it's difficult to be sure, but it looks like "tc checkfs pause" or "tc checkfs" or "tc base norestore checkfs" do not run the fs check - the re-boot part does however run.

Edit: would it be better to use e2fsck maybe?

Edit2: I just ran fsck manually - since the fs had been mounted a bunch of times, fsck ran for some considerable time so it definately did not run on boot with "tc checkfs"

[curaga]

Fsck is not forced currently, so it is only run if the filesystem is marked dirty (for instance after an unclean shutdown). Should it be forced?

[Juanito]

That was my point in edit2 above - the fs had been marked "dirty" as the mount count had gone over the limit. That aside, it would probably make sense to force fsck as the user presumeably wants it to run when using "checkfs"

[roberts]

I agree. Since it is be requested. I will force it.
Tested and working.

[curaga]

Happy new year

Around the new year's hacking I discovered we were using a statically compiled e2fsck. Why? It's 500kb of libc-overhead we have no need for (libc is always available when e2fsck is). Grab the dynamic version @ ibiblio. A new year's task for everyone: let's push TC to under 10mb

About tce-load: I think you meant

       for F in `cat "$APP".dep`; do
          DEPFILE="$FROMWHERE"/"$F"
-         if [ ! -f /usr/local/tce.installed/"$DEPFILE" ]; then
+         if [ ! -f /usr/local/tce.installed/"${F%.*}" ]; then
             install "$DEPFILE"
          fi   
       done

because with $DEPFILE it would have been /usr/local/tce.installed/ /path/to/ext/ .tcel, but now it's as it should /usr/local/tce.installed/ basename.

[roberts]

 @Curaga, nice!  That deserves a "B" not a grade but a cut,  rc9b

rc9b now posted! That's is for this year!

See you all next year!

Thanks and Happy New Year!

[Juanito]

ckfs now runs with "tc checkfs" - it might be worth echoing something to the screen so the user is reassured as to what's going on 

[roberts]

Fancy color messages will be in the final cut.