src/tests/parse_rss fails on 32 bits systems (but not i386)

[paulgevers]

I tried to improve the test coverage in the Debian package of liferea and uploaded what I though had the required ingredients to experimental. However, on 32 bit architectures, the parse_rss test fails because of a miss-match in the time. i386 passes (but that has weird floating point precision if I recall correctly).

See the results here:
https://buildd.debian.org/status/logs.php?pkg=liferea&ver=1.14.1-2

armel

"<item><title>i1</title><description>&lt;div xmlns=\"http://www.w3.org/1999/xhtml\"&gt;&lt;p&gt;D&lt;/p&gt;&lt;/div&gt;</description><source>http://localhost/item1.html</source><nr>0</nr><readStatus>0</readStatus><updateStatus>0</updateStatus><mark>0</mark><time>1678397817</time><sourceId/><sourceNr>0</sourceNr><attributes/></item>" == "<item><title>i1</title><description>&lt;div xmlns=\"http://www.w3.org/1999/xhtml\"&gt;&lt;p&gt;D&lt;/p&gt;&lt;/div&gt;</description><source>http://localhost/item1.html</source><nr>0</nr><readStatus>0</readStatus><updateStatus>0</updateStatus><mark>0</mark><time>5</time><sourceId/><sourceNr>0</sourceNr><attributes/></item>"

armhf

"<item><title>i1</title><description>&lt;div xmlns=\"http://www.w3.org/1999/xhtml\"&gt;&lt;p&gt;D&lt;/p&gt;&lt;/div&gt;</description><source>http://localhost/item1.html</source><nr>0</nr><readStatus>0</readStatus><updateStatus>0</updateStatus><mark>0</mark><time>1678397817</time><sourceId/><sourceNr>0</sourceNr><attributes/></item>" == "<item><title>i1</title><description>&lt;div xmlns=\"http://www.w3.org/1999/xhtml\"&gt;&lt;p&gt;D&lt;/p&gt;&lt;/div&gt;</description><source>http://localhost/item1.html</source><nr>0</nr><readStatus>0</readStatus><updateStatus>0</updateStatus><mark>0</mark><time>24068104</time><sourceId/><sourceNr>0</sourceNr><attributes/></item>"

mipsel

"<item><title>i1</title><description>&lt;div xmlns=\"http://www.w3.org/1999/xhtml\"&gt;&lt;p&gt;D&lt;/p&gt;&lt;/div&gt;</description><source>http://localhost/item1.html</source><nr>0</nr><readStatus>0</readStatus><updateStatus>0</updateStatus><mark>0</mark><time>1678397817</time><sourceId/><sourceNr>0</sourceNr><attributes/></item>" == "<item><title>i1</title><description>&lt;div xmlns=\"http://www.w3.org/1999/xhtml\"&gt;&lt;p&gt;D&lt;/p&gt;&lt;/div&gt;</description><source>http://localhost/item1.html</source><nr>0</nr><readStatus>0</readStatus><updateStatus>0</updateStatus><mark>0</mark><time>350198</time><sourceId/><sourceNr>0</sourceNr><attributes/></item>"

hppa

"<item><title>i1</title><description>&lt;div xmlns=\"http://www.w3.org/1999/xhtml\"&gt;&lt;p&gt;D&lt;/p&gt;&lt;/div&gt;</description><source>http://localhost/item1.html</source><nr>0</nr><readStatus>0</readStatus><updateStatus>0</updateStatus><mark>0</mark><time>1678397817</time><sourceId/><sourceNr>0</sourceNr><attributes/></item>" == "<item><title>i1</title><description>&lt;div xmlns=\"http://www.w3.org/1999/xhtml\"&gt;&lt;p&gt;D&lt;/p&gt;&lt;/div&gt;</description><source>http://localhost/item1.html</source><nr>0</nr><readStatus>0</readStatus><updateStatus>0</updateStatus><mark>0</mark><time>726728</time><sourceId/><sourceNr>0</sourceNr><attributes/></item>"

powerpc

"<item><title>i1</title><description>&lt;div xmlns=\"http://www.w3.org/1999/xhtml\"&gt;&lt;p&gt;D&lt;/p&gt;&lt;/div&gt;</description><source>http://localhost/item1.html</source><nr>0</nr><readStatus>0</readStatus><updateStatus>0</updateStatus><mark>0</mark><time>1678397817</time><sourceId/><sourceNr>0</sourceNr><attributes/></item>" == "<item><title>i1</title><description>&lt;div xmlns=\"http://www.w3.org/1999/xhtml\"&gt;&lt;p&gt;D&lt;/p&gt;&lt;/div&gt;</description><source>http://localhost/item1.html</source><nr>0</nr><readStatus>0</readStatus><updateStatus>0</updateStatus><mark>0</mark><time>5098</time><sourceId/><sourceNr>0</sourceNr><attributes/></item>"

[lwindolf]

Hah... I like when tests uncover bugs. Will take some time to work on it though...

[lwindolf]

@paulgevers I have some hope that #1233 fixes the issue. As I have no multi-arch build setup prepared (missing time) I'd like to ask you if you can test whether this helps.

[paulgevers]

I'm pretty sure I applied the patch but had this during test builds on armel

"<item><title>i1</title><description>&lt;div xmlns=\"http://www.w3.org/1999/xhtml\"&gt;&lt;p&gt;D&lt;/p&gt;&lt;/div&gt;</description><source>http://localhost/item1.html</source><nr>0</nr><readStatus>0</readStatus><updateStatus>0</updateStatus><mark>0</mark><time>1678397817</time><sourceId/><sourceNr>0</sourceNr><attributes/></item>" == "<item><title>i1</title><description>&lt;div xmlns=\"http://www.w3.org/1999/xhtml\"&gt;&lt;p&gt;D&lt;/p&gt;&lt;/div&gt;</description><source>http://localhost/item1.html</source><nr>0</nr><readStatus>0</readStatus><updateStatus>0</updateStatus><mark>0</mark><time>2</time><sourceId/><sourceNr>0</sourceNr><attributes/></item>"

armhf:

"<item><title>i1</title><description>&lt;div xmlns=\"http://www.w3.org/1999/xhtml\"&gt;&lt;p&gt;D&lt;/p&gt;&lt;/div&gt;</description><source>http://localhost/item1.html</source><nr>0</nr><readStatus>0</readStatus><updateStatus>0</updateStatus><mark>0</mark><time>1678397817</time><sourceId/><sourceNr>0</sourceNr><attributes/></item>" == "<item><title>i1</title><description>&lt;div xmlns=\"http://www.w3.org/1999/xhtml\"&gt;&lt;p&gt;D&lt;/p&gt;&lt;/div&gt;</description><source>http://localhost/item1.html</source><nr>0</nr><readStatus>0</readStatus><updateStatus>0</updateStatus><mark>0</mark><time>1074139144</time><sourceId/><sourceNr>0</sourceNr><attributes/></item>"

and mispel:

"<item><title>i1</title><description>&lt;div xmlns=\"http://www.w3.org/1999/xhtml\"&gt;&lt;p&gt;D&lt;/p&gt;&lt;/div&gt;</description><source>http://localhost/item1.html</source><nr>0</nr><readStatus>0</readStatus><updateStatus>0</updateStatus><mark>0</mark><time>1678397817</time><sourceId/><sourceNr>0</sourceNr><attributes/></item>" == "<item><title>i1</title><description>&lt;div xmlns=\"http://www.w3.org/1999/xhtml\"&gt;&lt;p&gt;D&lt;/p&gt;&lt;/div&gt;</description><source>http://localhost/item1.html</source><nr>0</nr><readStatus>0</readStatus><updateStatus>0</updateStatus><mark>0</mark><time>262345</time><sourceId/><sourceNr>0</sourceNr><attributes/></item>"

[paulgevers]

Version 1.15.1 now fails on the official Debian releases mips64el and s390x due to

./memcheck.sh parse_html favicon parse_date social
ERROR: memcheck reports problems for 'parse_html'!
==382210== Conditional jump or move depends on uninitialised value(s)
ERROR: memcheck reports problems for 'favicon'!
==382217== Conditional jump or move depends on uninitialised value(s)
ERROR: memcheck reports problems for 'parse_date'!
==382224== Conditional jump or move depends on uninitialised value(s)
ERROR: memcheck reports problems for 'social'!
==382231== Conditional jump or move depends on uninitialised value(s)

Of the unofficial releases, at least ppc64 is also affected.

See:
https://buildd.debian.org/status/fetch.php?pkg=liferea&arch=s390x&ver=1.15.1-1&stamp=1692516542&raw=0
https://buildd.debian.org/status/fetch.php?pkg=liferea&arch=mips64el&ver=1.15.1-1&stamp=1692519599&raw=0

[lwindolf]

@paulgevers In the current memcheck.sh script I've added support for supressions to be defined in memcheck.supp. If you want we can add suppressions for those platforms.