aboutsummaryrefslogtreecommitdiff
path: root/bpkg/rep-fetch.cxx
diff options
context:
space:
mode:
authorKaren Arutyunov <karen@codesynthesis.com>2018-02-12 17:22:50 +0300
committerKaren Arutyunov <karen@codesynthesis.com>2018-02-12 17:35:01 +0300
commitff06bfac111fe46f0b4453b19a7cfe2589644b87 (patch)
treef7a5646b1db002534eef857a232d30fb2795f08c /bpkg/rep-fetch.cxx
parent1ac50a35f4183cd7cd4bae0b310e20474a2d1f69 (diff)
Use abbreviated to 16 chars SHA256 fingerprint as certificate id
Diffstat (limited to 'bpkg/rep-fetch.cxx')
-rw-r--r--bpkg/rep-fetch.cxx26
1 files changed, 0 insertions, 26 deletions
diff --git a/bpkg/rep-fetch.cxx b/bpkg/rep-fetch.cxx
index 5566114..6ed62ab 100644
--- a/bpkg/rep-fetch.cxx
+++ b/bpkg/rep-fetch.cxx
@@ -99,32 +99,6 @@ namespace bpkg
//
// 6. Return repository and package manifest (certificate is NULL).
//
- // Notes:
- //
- // - Should we truncate sha256 hash? Maybe to 16 chars (this is what we
- // use for abbreviated git commit id in the version module). Also in
- // auth? Add abbreviated_string(size_t) to sha1 and sha256 classes?
- //
- // @@ If to truncate hash for auth, we would still need to store the full
- // fingerprint in the certificate object as rep-info needs it to print.
- // Leaving the certificate unchanged and truncating fingerprint on the
- // fly for the file naming seems wrong (good to have the certificate
- // file name to match the id). Probably it makes sense to make the
- // certificate as follows:
- //
- // class certificate
- // {
- // public:
- // string id; // SHA256 fingerprint truncated to 16 characters.
- //
- // string fingerprint; // Fingerprint canonical representation.
- // ...
- // };
- //
- // Yes, sounds good.
- //
- //
- //
if (conf != nullptr && conf->empty ())
conf = dir_exists (bpkg_dir) ? &current_dir : nullptr;