|
From: | Hubert Tarasiuk |
Subject: | Re: [Bug-wget] Metalink support |
Date: | Fri, 3 Jul 2015 13:54:43 +0200 |
User-agent: | Mozilla/5.0 (X11; Linux x86_64; rv:38.0) Gecko/20100101 Thunderbird/38.0.1 |
Hi Tim, W dniu 03.07.2015 o 10:17, Tim Ruehsen pisze: > Just a few minor points I stumbled upon when I tried to apply your patches :-) > > - they don't apply smoothly on latest master. Could you rebase your work on > current master ? > (What I did it is 'git am 00*' with your patches) That was my bad. It is now rebased. > > - I had to find out that I need libgpgme11-dev package (on Debian here), else > ./boostrap and/or ./configure would not work. Please add this information to > README.checkout. Both libmetalink and GPGME were supposed to be "weak" dependencies. How about copying GPGME's m4 macro to Wget's m4 directory? If I add it to the Wget repository, the scripts will not fail without GPGME installed. > > - still libmetalink was missing and a 'make check' failed with > ../src/libunittest.a(libunittest_a-test.o): In function `all_tests': > /usr/oms/src/wget/src/test.c:51: undefined reference to `test_find_key_value' > /usr/oms/src/wget/src/test.c:52: undefined reference to `test_find_key_values' > /usr/oms/src/wget/src/test.c:53: undefined reference to `test_has_key' > Please add a #ifdef to src/test.c and also mention libmetalink in > README.checkout. I have repaired this problem. I have also conditionally disabled the Metalink Python tests (whenever Metalink support is not compiled). Thus, Wget should work fine without neither of them. (Both make and make-check.) Should I list these dependencies as optional in the README.checkout? > > It seems that libmetalink hasn't been packaged by Debian yet. I'll come back > when I have libmetalink installed... I see. The 0.13 version is not yet available on ArchLinux either, so I am compiling it manually at this moment. > (I'll find some time at the weekend). Please note that I will not be able to commit changes after tomorrow night (the schedule of my GSOC is altered). However, if you come up with any more issues before that time, I will do my best to resolve them. Thank you, Hubert > > Regards, Tim > > On Friday 03 July 2015 00:56:21 Hubert Tarasiuk wrote: >> I have adjusted some comments to reflect the current libmetalink situation. >> >> Moreover, I added another option: --preferred-location=<location> that >> can be used to give priority to certain Metalink resources based on >> their location. >> >> These changes are presented in three additional commits (10, 11, 12) for >> easier review (against previous patches). They should be probably >> squashed before merging: >> squash 1,9,10,11 into one >> squash 5,12 into one >
0001-Metalink-support.patch
Description: Text Data
0002-Start-HTTP-test-only-when-calling-begin.patch
Description: Text Data
0003-Test-case-for-Metalink-in-XML.patch
Description: Text Data
0004-Support-multiple-headers-with-same-name-in-Python-te.patch
Description: Text Data
0005-Test-case-for-Metalink-over-HTTP.patch
Description: Text Data
0006-Unit-test-for-find_key_value.patch
Description: Text Data
0007-Unit-test-for-has_key.patch
Description: Text Data
0008-Unit-test-for-find_key_values.patch
Description: Text Data
0009-Move-some-Metalink-related-code-from-http.c-to-metal.patch
Description: Text Data
0010-Support-at-most-one-file-signature.-Adapt-comments-t.patch
Description: Text Data
0011-Geolocation-support-for-Metalink-resources.patch
Description: Text Data
0012-Test-preferred-location-in-Metalink-over-HTTP-test-c.patch
Description: Text Data
signature.asc
Description: OpenPGP digital signature
[Prev in Thread] | Current Thread | [Next in Thread] |