gemma-wrapper: update

1 files changed, 44 insertions, 40 deletions

diff --git a/gn/packages/gemma.scm b/gn/packages/gemma.scm
index 9b44759..93bcbdd 100644
--- a/gn/packages/gemma.scm
+++ b/gn/packages/gemma.scm
@@ -23,6 +23,7 @@
   #:use-module (gnu packages time)
   #:use-module (gnu packages web)
   #:use-module (gnu packages ruby-check)
+  #:use-module (gnu packages ruby-xyz)
   #:use-module (gn packages shell)
   #:use-module (srfi srfi-1))
 
@@ -110,24 +111,15 @@ genome-wide association studies (GWAS).")
 
 
 (define-public gemma-wrapper
-  ;; Source: upstream v0.99.7 (commit 48c18a6).  The published rubygem
-  ;; ships only bin/gemma-wrapper + lib/lock.rb, no Rakefile and no
-  ;; test data; the git tag includes everything we need for the LOCO
-  ;; regression test.
-  ;;
-  ;; TODO: a Rakefile fix that warms the GWA cache before asserting
-  ;; cache_hit (so the upstream tests pass on a fresh /tmp) has been
-  ;; cherry-picked onto a v0.99.7-rakefile-fix branch as commit
-  ;; a070594f05f2ed91f8099a4c2d6fa5df843c2584 (sha256 nar
-  ;; 056kcs39g10n5bzyi4l62g9625mnwcznaxk167dydfja0ras8chv).  Once
-  ;; that commit lands on a public branch, switch the (commit ...)
-  ;; below to a070594 and drop the Rakefile substitute* in the
-  ;; set-gemma-path phase.
-  (let ((commit "48c18a6c1ae55da51e62b2d7308b959aefc3a683")
-        (revision "0"))
+  ;; Source: upstream master (commit 3a9286c, version 1.00-pre1).  The
+  ;; published rubygem ships only bin/gemma-wrapper + lib/lock.rb, no
+  ;; Rakefile and no test data; the git tag includes everything we
+  ;; need for the LOCO regression test.
+  (let ((commit "3a9286c92ebe8d177fb0ca3b776aba1ddfce9904")
+        (revision "1"))
     (package
       (name "gemma-wrapper")
-      (version (git-version "0.99.7" revision commit))
+      (version (git-version "1.00-pre1" revision commit))
       (source
        (origin
          (method git-fetch)
@@ -136,19 +128,21 @@ genome-wide association studies (GWAS).")
                (commit commit)))
          (file-name (git-file-name name version))
          (sha256
-          (base32 "17jlj5pl184wvh5s7gyv3qi87iw8f1p6yd2psgmqa7nkan744mgb"))))
+          (base32 "1hfj4cr3l21k6sk308d2gvwlky2szyl1ziv364iv3q93rhjks59d"))))
       (build-system ruby-build-system)
       (native-inputs (list ruby-rake))
-      (inputs (list gemma-gn2))
       (propagated-inputs
        ;; bin/gemma-wrapper shells out to all of these; propagate
        ;; them so `guix shell gemma-wrapper` is a complete runtime.
-       (list parallel      ;; orchestrates per-chromosome and permutation jobs
-             coreutils      ;; uses cat, env, rm
-             tar            ;; archives GEMMA's per-run outputs as .tar.xz
-             xz             ;; tar -J needs xz on PATH
-             time           ;; bin/gemma-wrapper invokes `time -v gemma ...`
-             pfff))         ;; fast file fingerprint for inputs >100KB
+       (list parallel       ;; orchestrates per-chromosome and permutation jobs
+             coreutils       ;; uses cat, env, rm
+             gemma-gn2
+             tar             ;; archives GEMMA's per-run outputs as .tar.xz
+             xz              ;; tar -J needs xz on PATH
+             time            ;; bin/gemma-wrapper invokes `time -v gemma ...`
+             pfff            ;; fast file fingerprint for inputs >100KB
+             ruby-rdf        ;; gemspec runtime dep (RDF helpers in bin/)
+             ruby-rdf-vocab));; gemspec runtime dep (RDF helpers in bin/)
       (arguments
        (list
         #:phases
@@ -160,22 +154,30 @@ genome-wide association studies (GWAS).")
                    (string-append "gemma_command = '"
                                   #$(file-append gemma-gn2 "/bin/gemma")
                                   "'"))
-                  ;; v0.99.7 bug: `"..."+options[:trait]+"..."` crashes
-                  ;; with TypeError when --trait isn't passed (the
-                  ;; bundled Rakefile test never sets it).  Switch to
-                  ;; string interpolation which renders nil as "".
+                  ;; v0.99.7/1.00-pre1 bug: `"..."+options[:trait]+"..."`
+                  ;; crashes with TypeError when --trait isn't passed
+                  ;; (the bundled Rakefile test never sets it).  Switch
+                  ;; to string interpolation which renders nil as "".
                   (("\"https://genenetwork.org/show_trait\\?trait_id=\"\\+options\\[:trait\\]\\+\"&dataset=\"\\+options\\[:name\\]")
                    "\"https://genenetwork.org/show_trait?trait_id=#{options[:trait]}&dataset=#{options[:name]}\""))
-                ;; Temporary: the Rakefile's GWA cache_hit assertions
-                ;; presume warm /tmp; v0.99.7 itself is broken on a
-                ;; fresh sandbox.  A proper fix (cherry-picked onto
-                ;; the v0.99.7-rakefile-fix branch as a070594) drops
-                ;; this block; until upstream pushes that branch we
-                ;; replicate the same edit here.  Remove once the
-                ;; source commit above is bumped to a070594.
+                ;; ruby-lmdb is not yet packaged in Guix; strip it
+                ;; from the gemspec so gem activation can succeed.
+                ;; The *mdb* helpers in bin/ will still abort at
+                ;; `require 'lmdb'` until ruby-lmdb is packaged.
+                (substitute* "gemma-wrapper.gemspec"
+                  (("s\\.add_runtime_dependency 'lmdb'[^\n]*\n") ""))
+                ;; The Rakefile asserts pre-1.00-pre1 K/GWA SHA1
+                ;; baselines that don't match the new hash algorithm
+                ;; in commit 3a9286c, and asserts `"cache_hit":true`
+                ;; on JSON outputs that no longer carry that field
+                ;; (1.00-pre1 restructured the record into meta/archive
+                ;; entries).  Drop both classes of assertion; the
+                ;; errno=0 + "Test failed" exit-code checks still
+                ;; gate the LOCO pipeline.  Remove once the Rakefile
+                ;; baselines are refreshed upstream.
                 (substitute* "Rakefile"
-                  (("fail \"Expected cache hit in #\\{gwa0\\}\"[^\n]*\n") "")
-                  (("fail \"Expected cache hit in #\\{gwa2\\}\"[^\n]*\n") ""))))
+                  (("fail \"Wrong Hash in #\\{[^}]+\\}\"[^\n]*\n") "")
+                  (("fail \"Expected cache hit in #\\{[^}]+\\}\"[^\n]*\n") ""))))
             ;; v0.99.7 ships a working `rake test`: it runs the LOCO
             ;; pipeline (non-LOCO -gk, LOCO -gk chr1-4, GWA with
             ;; cache hits) and asserts the expected SHA1 hashes in
@@ -188,13 +190,15 @@ genome-wide association studies (GWAS).")
                   ;; above.  parallel must also be on PATH for the
                   ;; LOCO step's fork-out.
                   ;; gemma-wrapper shells out to `time -v gemma ...`
-                  ;; (GNU time, not the bash builtin) and to `parallel`
-                  ;; for the LOCO fork-out.  Both need to be on PATH
-                  ;; during the test invocation.
+                  ;; (GNU time, not the bash builtin), to `parallel`
+                  ;; for the LOCO fork-out, and (since 1.00-pre1) to
+                  ;; `pfff` for input fingerprinting.  All need to be
+                  ;; on PATH during the test invocation.
                   (setenv "PATH"
                           (string-append
                            #$(file-append parallel "/bin") ":"
                            #$(file-append time "/bin") ":"
+                           #$(file-append pfff "/bin") ":"
                            (or (getenv "PATH") "")))
                   ;; lib/lock.rb writes "$HOME/.<hash>.lck" lock files;
                   ;; the Guix sandbox sets HOME=/homeless-shelter which