Hi,
http://ftp-master.debian.org/testing/testing_outdate.txt has bothered me
since I discovered it a few times. It shows m68k has 22 out of date with
mipsel and hppa at 6. Most are at 0 or 1.

Please consider building/uploading clustalw 1.83-1.2

What's going on with etch-secure? See
http://etch-secure.buildd.net/index-m68k.html that things are claimed
but seemingly nothing's been built for a long time. What is this? An
unembargoed pre-release version of the security buildd setup?

Where's tpu, spu buildd information? I'm guessing security uploads are
embargoed, but proposed updates shouldn't be security updates right?

Some ftbfs problems (add to reasons:
http://buildd.debian.org/fetch.cgi?&pkg=ctn&ver=3.0.6-10&arch=m68k&stamp=1171341799&file=log
[...]
ln -s ../facilities/dicom/intsizes.h /build/buildd/ctn-3.0.6/libsrc
ln: creating symbolic link `/build/buildd/ctn-3.0.6/libsrc/intsizes.h' to 
`../facilities/dicom/intsizes.h': File exists
make[3]: *** [links] Error 1
[...]

http://buildd.debian.org/fetch.cgi?&pkg=fcitx&ver=1%3A3.4.3-1&arch=m68k&stamp=1171495231&file=log
[...]
m68k-linux-gnu-gcc  -Wall -g -O2 -INONE -INONE -I/usr/include/freetype2 
-D_USE_XFT -Wall   -o mb2org  mb2org.o ../src/pyParser.o ../src/pyMapTable.o 
../src/PYFA.o ../src/sp.o pyTools.o -LNONE -lXft -lfontconfig   -lX11 
./txt2mb ./../data/cj.txt ../data/cj.mb

Reading 6726 records.

./txt2mb ./../data/erbi.txt ../data/erbi.mb

Reading 82911 records.

./txt2mb ./../data/wbx.txt ../data/wbx.mb

Reading 68782 records.

./createPYMB ./../data/gbkpy.org ./../data/pyPhrase.org
make[3]: *** [all] Segmentation fault
[...]


I'm occasionally looking through the testing script output then:
1 checking to see if that version failed to build or was never built
 1a if it was never built, request it be built
 1b if it was built, bug someone to upload and install?
 1c if it failed to build, check if it ftbfs on other architectures
  1ci if it didn't ftbfs on other archs, request consider requesting
      rebuild
  1cii if it did ftbfs on other archs, almost automatically file a bug

So some of this might be able to be automated and save some future
porter time. Also adding links between some of these pages would save
time. I guess later I'll:
- bug smarenka to find out if his pages are automatically generated
- bug d-release (probably aj) about links from the testing output to 
  buildd.d.o.
- 1b sounds like it could generate an email for the porters list or 
  possibly ftpmaster if it has been uploaded? Who works on this
  software? I'll look for someone to discuss this with.
- find buildd.d.o person to help mark ftbfs that's only on one arch? I
  guess igloo does this?
- Discuss 1ci with 1b people. Certainly this kind of failure should be
  flagged somehow.
- Discuss 1cii on debian-devel. Perhaps a script to check buildd
  failures and automatically file bugs when a package files on multiple
  archs? The script may also be able to close such bugs when they do get
  fixed. Maybe buildd maybe-failed should always trigger filing a bug
  and maybe-success could be tied to it and automatically close such a
  bug (with appropriate version tags).

Fwiw, I'd like it if all of buildd.net's targets were being built. If
that means more buildd's, then why not? backports shows no buildds.
non-free may need another buildd. sarge-volatile is built, but may be
problematic later (where's ekg's libreadline4 though? dep-wait since 2005?).
I've heard some people talk about not building testing, or not releasing
with etch. Isn't the port just about ready for etch? Most of the
remaining ftbfs could be made not-for-us (nfu) on m68k right? E.g.
mozart, r-base... vorlon or aj seemed to suggest once that porters
should mark more packages as nfu (although I'm sure they want to avoid
dependency problems).

I supposed I've rambled on a bit, but then I'm hoping if my comments
don't help the port much directly, it'll help it indirectly by helping
me learn the buildd process better.
Thanks,

     Drew Daniels
Resume: http://www.boxheap.net/ddaniels/resume.html


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Reply via email to