booBot

Here are some photos of the tools I use.

Посмотреть: (1280x960) 356228 байт

Посмотреть: (1280x960) 469618 байт

Посмотреть: (1280x960) 422731 байт

Here is the picture of a 16MB Flash (Intel E28F128J3A-150) chip on a PCB.

Посмотреть: (800x600) 250021 байт

Here is a small video telling how to manage TSOP chips with just an ordinary tool.

This is an attempt to re-invent the Linksys's gen_mc utility to generate the SSL certificates for RT31P2 and|or PAP2|SPA class of devices.

Included is the renamed (as usual here) tar.gz C source file.
Known to compile on QNX4 and LINUX.

The APR is really cool stuff - it has it's own implementation of the apr_snprintf() so that it almost works even without the underlaying int64-aware printf()|scanf().

Of the four failing tests in the ./test/testfmt.c three were actually from the QNX4's native snprintf().
The fourth failure is credited to the apr_snprintf() which handles well any unsigned value and breaks on the signed one larger or equal to 0x80000000. May be it is not even the apr_snprintf()'s fault again but my toolchain's...

I'm looking into the issue.

I'm desperately trying to create a usable version of the clib with the int64-aware printf()|scanf().

I have three libraries to fool around with:
1) the original Watcom C v10.6B clib3r.lib;
2) the original Watcom C v11.0beta clib3r.lib;
3) and the half-baked Mario's clib3r.lib from the OpenWATCOM v1.6RC1;

My current toolchain is the combination of OW v1.6RC1 compiler and the WC v11.0beta clib3r.lib.
Unless I need to printf() things as %Lsomething - it works fine.

I did try to borrow the missing (in the OW's clib) objects from the WC's v11.0 clib. Only the very "HelloWorld"-like programs compile and actually work. The APR v1.2.8 or something really serious - does not...
May be I should try to go the other way around - to replace the old printf()|scanf() in the Watcom C v11.0beta's clib, but my first try SIGSEGVs immediately...

Let's hope Mario will come to the resque again...

For now, here is the diff to introduce the QNX4-like prefix|path|filename handling into the APR.

The only outstanding issue is that APR wants the strftime() to have the %R directive which is similar to the existing %T but does not include the seconds field. This needs hacking the clib, I believe.
Unfortunately, Mario is very busy lately...

Untill Mario releases his port of OpenWATCOM v1.6 with the fully usable clib3r.lib I will not be able to check for sure if I ported the APR v1.2.8 properly.

So far I was able to make APR to support FLEET.

(There are 4 tests failing, all related to printing the int64 values and similar directives. All other tests pass OK.)

No major changes are needed, just a cleverly selected features to keep enabled in the ./include/apr.h - like disabling the mmap() and friends, disabling the shared memory usage.

For now I move on to try porting the APR-UTIL v1.2.8.

My friends just came back from Ceylon and it suddenly struck me: a user, in his infinite wisdom, may choose a TZ setting where the DST adjustment is not exactly an hour and not actually positive...

So, please redownload the library where the TZ is handled properly: the extern time_t __dst_adjust value is used (which follows the TZ rules) rather than the hardcoded +3600.

The neon v0.26.3 library is compiled with Watcom C v10.6B, linked with:
OpenSSL v0.9.8d
zlib v1.2.3
expat v2.0.0

I apologize for inconveniences caused! :)

Because of no support for the long long in the Watcom C v10.6B I'm able to do just the very basic attempts with Mario Charest's port of OpenWATCOM v1.6RC1.

Currently I dig the ./file-io/unix/filepath.c to bring the FLEET-awareness into the APR library.

I keep my eye on Mario's blog (http://zinformatic.blogspot.com/) for the OW news and porting progress.

Now all problems with the neon v0.26.3 on QNX4 are solved.

I'm moving on to port the APR v1.2.8 and the APR-UTIL v1.2.8.

(the package and the diff are updated on 01feb2007, see above)

Here is the diff to the neon v0.26.3.

I can't figure out what's wrong with the 25th test of the socket test suite. (the large_writes test fails with the "(-3) broken pipe" error)

Any help is very welcome!

(Download the attached file, remove the "mp3" file extention and apply.)

Done so far:
1) ported expat v2.0.0 library
2) porting neon v0.26.3 library (almost done, just one test fails)