Arch Linux User Repository

@lightdot

Thanks for the hint which helps me understand the reason. If due to some reason, the download files got corrupted, or updated but remain same version, pkgbuild will not re-download (check existence) but only compare hash, then reports "not pass the validity check".

A clean download and makepkg can always work, however I thought PKGBUILD should provide an option to validate hash for existing file, if not matching, always re-download, so that sha256sums and latest downloads always pair, instead of simply compare sha256sums with existing file's.

@Ivellios,

after seeing your report, I was certain the patch got silently updated and thus has a new SHA256 digest. It happened in the past.

But when downloading the patch file to check, I still get the file with the current digest, as listed in the PKGBUILD.

The actual content in mine is dated 2023-12-24 (with two files having a 2023-12-25 date and others 2023-12-24).

The upstream publishes the maildir patch digests here: https://alpineapp.email/alpine/md5/maildir.html

Both the current digest and the one you've reported are there. The current one is listed last. I'm assuming that's the most recent one, but can't be sure.

I'm also not sure why you got served a different patch file. Some kind of caching or download generation on-demand perhaps... Odd.

==> Validating source files with sha256sums...
    alpine-2.26.tar.xz ... Passed
    maildir-2.26.patch.gz ... FAILED
    fancy-2.26.patch.gz ... Passed
    fillpara-2.26.patch.gz ... Passed
    compose-2.26.patch.gz ... Passed
    longurl-2.26.patch.gz ... Passed
$ sha256sum maildir-2.26.patch.gz
01020ca0cec24793a39d5ab72c0beefc0cfbb92d11348b5bd5bf7bfbcb203f63  maildir-2.26.patch.gz

$ cat PKGBUILD | grep -A 5 -e sha256sums
sha256sums=('c0779c2be6c47d30554854a3e14ef5e36539502b331068851329275898a9baba'
        '57808418b02a0e1cb826940068aa10eaed827b2b23609f05fd2015ec92043d37'
        'c178459dd885e4caa32640e5ce63c689ec4752a1e039ddb149e034b935fe5181'
        '960bb9656353529964e86f0f782bf032f9aae36af9493ed7f2c8ccfd5695330f'
        '013d31d95dbf6e31c1ebfdcb745481cf31fd0df466b81e077a7538dcfe75a9f4'
        'b29b4b6f8986c4c07ce2db21b6ff27341b4f7c0844c3dee8c1c63b0536fc88b2')

Adding to configure phase:

CFLAGS="-Wno-error=incompatible-pointer-types"

allows alpine to compile. I believe it is OK in this case to work around this, albeit optimally upstream should fix this in the code.

diff --git a/PKGBUILD b/PKGBUILD
index 50fe1ec..8df4cfd 100644
--- a/PKGBUILD
+++ b/PKGBUILD
@@ -46,6 +46,7 @@ build() {
        patch -p1 -i "${srcdir}/longurl-${pkgver}.patch"

        # Configure Alpine
+       CFLAGS="-Wno-error=incompatible-pointer-types" \
        LIBS+="-lpam -lkrb5 -lcrypto" ./configure --prefix=/usr \
        --with-passfile=.alpine.passfile --without-tcl --disable-shared \
        --with-system-pinerc=/etc/${pkgname}.d/pine.conf \

can't find https://alpineapp.email/alpine/release/src/alpine-2.26.tar.xz

Confirm Wild_Penguin and cmcc as well.

Indeed does not compile here either, I get the same error as cmcc.

After upgrading gcc to version 14.1.1, it seems alpine doesn't compile anymore. Compilation stops when trying to compile adrbklib.c. The error I get is

gcc -DHAVE_CONFIG_H   -I../include -I../include -I/usr/include  -g  -march=x86-64 -mtune=native
-march=native -O3 -pipe -fno-plt -fexceptions -Wp,-D_FORTIFY_SOURCE=2 -Wformat -Werror=format-security
-fstack-clash-protection -fcf-protection -DOPENSSL_1_1_0 -DOPENSSL_API_COMPAT=0x10100000L -MT adrbklib.o -MD -MP -MF
.deps/adrbklib.Tpo -c -o adrbklib.o adrbklib.c
adrbklib.c: In function ‘sort_addr_list’:
adrbklib.c:4705:64: error: passing argument 4 of ‘qsort’ from incompatible pointer type [-Wincompatible-pointer-types]
 4705 |     qsort((qsort_t *)list, (size_t)(p - list), sizeof(char *), cmp_addr);
      |                                                                ^~~~~~~~
      |                                                                |
      |                                                                int (*)(const char *, const char *)
In file included from ../include/system.h:42,
                 from ../pith/headers.h:25,
                 from adrbklib.c:14:
/usr/include/stdlib.h:971:34: note: expected ‘__compar_fn_t’ {aka ‘int (*)(const void *, const void *)’} but argument is of type ‘int (*)(const char *, const char *)’
  971 |                    __compar_fn_t __compar) __nonnull ((1, 4));
      |                    ~~~~~~~~~~~~~~^~~~~~~~
adrbklib.c: In function ‘adrbk_sort’:
adrbklib.c:4766:77: error: passing argument 4 of ‘qsort’ from incompatible pointer type [-Wincompatible-pointer-types]
 4765 |             (ab->sort_rule == AB_SORT_RULE_FULL_LISTS) ?
      |             ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~                     
 4766 |                                                 cmp_cntr_by_full_lists_last :
      |                                                 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~^
      |                                                                             |
      |                                                                             int (*)(const char *, const char *)
 4767 |             (ab->sort_rule == AB_SORT_RULE_FULL) ?
      |             ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~                           
 4768 |                                                 cmp_cntr_by_full :
      |                                                 ~~~~~~~~~~~~~~~~~~           
 4769 |             (ab->sort_rule == AB_SORT_RULE_NICK_LISTS) ?
      |             ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~                     
 4770 |                                                 cmp_cntr_by_nick_lists_last :
      |                                                 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 4771 |             /* (ab->sort_rule == AB_SORT_RULE_NICK) */
      |             ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~                       
 4772 |                                                 cmp_cntr_by_nick);
      |                                                 ~~~~~~~~~~~~~~~~             
/usr/include/stdlib.h:971:34: note: expected ‘__compar_fn_t’ {aka ‘int (*)(const void *, const void *)’} but argument is of type ‘int (*)(const char *, const char *)’
  971 |                    __compar_fn_t __compar) __nonnull ((1, 4));
      |                    ~~~~~~~~~~~~~~^~~~~~~~
make[4]: *** [Makefile:589: adrbklib.o] Error 1
make[4]: Leaving directory '/home/xyz/.cache/yay/alpine/src/alpine-2.26/pith'
make[3]: *** [Makefile:622: all-recursive] Error 1
make[3]: Leaving directory '/home/xyz/.cache/yay/alpine/src/alpine-2.26/pith'
make[2]: *** [Makefile:460: all] Error 2
make[2]: Leaving directory '/home/xyz/.cache/yay/alpine/src/alpine-2.26/pith'
make[1]: *** [Makefile:547: all-recursive] Error 1
make[1]: Leaving directory '/home/xyz/.cache/yay/alpine/src/alpine-2.26'
make: *** [Makefile:436: all] Error 2

@RAMChYLD, according to https://alpineapp.email/alpine/md5/maildir.html the patch was updated without changing the version of either the patch or the package.

I didn't notice the update, so the current PKGBUILD doesn't reflect the newest code. I'll update soon.

In the meantime, as long as the current SHA256SUM published by the author corresponds with the one of the file you've downloaded, I'd trust it.

@simona, sorry for not responding sooner. I cannot reproduce this, though.

Something's wrong with the maildir-2.26 patch

==> Validating source files with sha256sums...
    alpine-2.26.tar.xz ... Passed
    maildir-2.26.patch.gz ... FAILED
    fancy-2.26.patch.gz ... Passed
    fillpara-2.26.patch.gz ... Passed
    compose-2.26.patch.gz ... Passed
    longurl-2.26.patch.gz ... Passed
==> ERROR: One or more files did not pass the validity check!
 -> error downloading sources: /home/leecy/.cache/yay/alpine 
         context: exit status 1 

Can you please confirm that the patch is not tainted or corrupted in any way?