summaryrefslogtreecommitdiffstats
path: root/README
diff options
context:
space:
mode:
authorDaniel Baumann <daniel.baumann@progress-linux.org>2024-04-15 18:12:25 +0000
committerDaniel Baumann <daniel.baumann@progress-linux.org>2024-04-15 18:12:25 +0000
commit0013fdf36b7348479f57d989a014bdd9e8a656f1 (patch)
treecc90b6104296a98c953a704d6b0eab0ca5542010 /README
parentInitial commit. (diff)
downloadisa-support-0013fdf36b7348479f57d989a014bdd9e8a656f1.tar.xz
isa-support-0013fdf36b7348479f57d989a014bdd9e8a656f1.zip
Adding upstream version 21.upstream/21
Signed-off-by: Daniel Baumann <daniel.baumann@progress-linux.org>
Diffstat (limited to 'README')
-rw-r--r--README21
1 files changed, 21 insertions, 0 deletions
diff --git a/README b/README
new file mode 100644
index 0000000..629a1bb
--- /dev/null
+++ b/README
@@ -0,0 +1,21 @@
+This package ensures that the machine has required CPU facilities; users who
+attempt installation on an inadequate processor will receive an error message
+in preinst. This is useful if there's no way to make your package support
+the base ISA for the architecture.
+
+While in general packages are required to at least try, there are situations
+when baseline support is infeasible or a pure waste of time. No one is
+going to run scientific number-crunching software on a pre-SSE2/SSE3
+machine. Or upstream code may use such extensions unconditionally, and
+you don't have the resources to fix that yourself.
+
+You should also consider how likely other software is to use your package.
+A fast strong hash is something with many uses, big and small, so it's a
+priority to port libraries like libhighwayhash to all CPUs. On the other
+hand, a niche piece of software is unlikely to be repurposed for other uses.
+
+
+The check is skipped in two cases:
+* when the target architecture is not executable at all (an ARM SoC mounted
+ into an amd64 machine, etc)
+* when ISA_IGNORE environment variable has a non-empty value