1 |
On Sat, 27 Sep 2008 12:58:40 -0400 |
2 |
Kirk Lowery wrote: |
3 |
|
4 |
> On Fri, Sep 26, 2008 at 2:47 AM, Neil Bothwick <neil@××××××××××.uk> |
5 |
> wrote: |
6 |
> > On Thu, 25 Sep 2008 17:59:35 -0400, Kirk Lowery wrote: |
7 |
> > |
8 |
> >> During upgrading today, I inadvertently allowed 2.3 to be deleted. |
9 |
> >> I looked in portage, but it is apparently gone. |
10 |
> > |
11 |
> > Nothing is ever truly gone from portage |
12 |
> > |
13 |
> > http://sources.gentoo.org/viewcvs.py/gentoo-x86/dev-lang/python/?hideattic=0 |
14 |
> > |
15 |
> > goes back as far as python-1/5/2 |
16 |
> |
17 |
> Thanks for the answers. I thought I'd report my experience here for |
18 |
> the knowledgebase: |
19 |
> |
20 |
> I went back and downloaded not only the latest 2.3.6 ebuild into my |
21 |
> local overlay, and all the attendant files (including important |
22 |
> patches) in the "files" directory. In order to be able to emerge the |
23 |
> ebuild file, I had to recreate the digest and manifest files. Don't |
24 |
> know why the checksums were off, but they were. The solution was |
25 |
> simple: |
26 |
> |
27 |
> # ebuild python-2.3.6-r6.ebuild digest |
28 |
> |
29 |
> took care of it. When I tried to restart my legacy software I got a |
30 |
> python error: "ImportError: No module named thread". Sure enough, the |
31 |
> equery use feature told me "-threads". So I overrode that in |
32 |
> package.use. Nope, no change. Looking at the compile log, I saw this: |
33 |
> |
34 |
> ./configure --prefix=/usr --host=x86_64-pc-linux-gnu |
35 |
> --mandir=/usr/share/man --infodir=/usr/share/info --datadir=/usr/share |
36 |
> --sysconfdir=/etc --localstatedir=/var/lib --with-fpectl |
37 |
> --enable-shared --enable-ipv6 --infodir=${prefix}/share/info |
38 |
> --mandir=${prefix}/share/man --with-libc= --enable-unicode=ucs4 |
39 |
> --with-threads --without-threads --with-threads --libdir=/usr/lib64 |
40 |
> --build=x86_64-pc-linux-gnu |
41 |
> |
42 |
> Configure has both --with- and --without-threads!!?? |
43 |
> |
44 |
> Nothing I did seemed to make a difference. So I traced where in the |
45 |
> ebuild this might be coming from: |
46 |
> |
47 |
> use threads \ |
48 |
> && myconf="${myconf} --with-threads" \ |
49 |
> && myconf="${myconf} --without-threads" |
50 |
> |
51 |
> I still don't know why (and would appreciate any knowledgeable person |
52 |
> commenting here), but I simply commented out the third line above, |
53 |
> re-emerged, and viola! my legacy software is up and running. |
54 |
> |
55 |
> Kirk |
56 |
|
57 |
An interesting defect to find, and a good job of detective work! |
58 |
|
59 |
Out of curiosity, I looked at the python ebuilds on my workstation. |
60 |
|
61 |
Up through python-2.4.4-r6, they have: |
62 |
|
63 |
use nothreads \ |
64 |
&& myconf="${myconf} --without-threads" \ |
65 |
|| myconf="${myconf} --with-threads" |
66 |
|
67 |
From python-2.4.4-r14 onwards they have: |
68 |
|
69 |
use threads \ |
70 |
&& myconf="${myconf} --with-threads" \ |
71 |
|| myconf="${myconf} --without-threads" |
72 |
|
73 |
Evidently the 2.3.6-r6 ebuild has the newer "use threads" test, but |
74 |
with the minor (but fatal) operator flaw you found -- "&&" which |
75 |
should be "||". |
76 |
|
77 |
Hope this is of interest, |
78 |
|
79 |
David |