Skip to content

Commit 1b7646e

Browse files
author
Rahul Iyer
committed
Upgrade: Add script to upgrade to v1.3 from v1.0, v1.1, v1.2
Pivotal Tracker: 57615140 Changes: - Added all changelist files to upgrade to v1.3 - Updated script to new upgrade - Fixed issues around whitespace between arguments - Release notes for v1.3 - Updated version numbers - Update PGXN versioning style to accomodate requirements
1 parent e6b44cb commit 1b7646e

File tree

14 files changed

+710
-385
lines changed

14 files changed

+710
-385
lines changed

‎ReleaseNotes.txt‎

Lines changed: 36 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -8,6 +8,42 @@ A complete list of changes for each release can be obtained by viewing the git
88
commit history located at https://github.com/madlib/madlib/commits/master.
99

1010
Current list of bugs and issues can be found at http://jira.madlib.net.
11+
--------------------------------------------------------------------------------
12+
MADlib v1.3
13+
14+
Release Date: 2013-October-03
15+
16+
New Features:
17+
* Cox Proportional Hazards:
18+
- Added stratification support for Cox PH models. Stratification is used as
19+
shorthand for building a Cox model that allows for more than one stratum,
20+
and hence, allows for more than one baseline hazard function.
21+
Stratification provides two pieces of key, flexible functionality for the
22+
end user of Cox models:
23+
-- Allows a categorical variable Z to be appropriately accounted for in
24+
the model without estimating its predictive impact on the response
25+
variable.
26+
-- Categorical variable Z is predictive/associated with the response
27+
variable, but Z may not satisfy the proportional hazards assumption
28+
- Added a new function (cox_zph) that tests the proportional hazards
29+
assumption of a Cox model. This allows the user to build Cox models and then
30+
verify the relevance of the model.
31+
* NULL Handling:
32+
- Modified behavior of linear and logistic regression to 'omit' rows
33+
containing NULL values for any of the dependent and independent variables.
34+
The number of rows skipped is provided as part of the output table.
35+
36+
Deprecated functions:
37+
- Cox Proportional Hazard function has been renamed to 'coxph_train'.
38+
Old function names ('cox_prop_hazards' and 'cox_prop_hazards_regr')
39+
have been deprecated, and will be removed in the next major version update.
40+
- The aggregate form of linear regression ('linregr') has been deprecated.
41+
The stored-procedure form ('linregr_train') should be used instead.
42+
43+
Bug Fixes:
44+
- Fixed a memory leak in the Apriori algorithm.
45+
46+
1147
--------------------------------------------------------------------------------
1248
MADlib v1.2
1349

‎deploy/PGXN/CMakeLists.txt‎

Lines changed: 3 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -3,8 +3,10 @@
33
# ------------------------------------------------------------------------------
44

55
set(MADLIB_PGXN_RELEASE_NUMBER 1)
6+
# set(MADLIB_PGXN_VERSION_STR
7+
# "${MADLIB_VERSION_MAJOR}.${MADLIB_VERSION_MINOR}.${MADLIB_VERSION_PATCH}release${MADLIB_PGXN_RELEASE_NUMBER}
68
set(MADLIB_PGXN_VERSION_STR
7-
"${MADLIB_VERSION_MAJOR}.${MADLIB_VERSION_MINOR}.${MADLIB_VERSION_PATCH}release${MADLIB_PGXN_RELEASE_NUMBER}")
9+
"${MADLIB_VERSION_MAJOR}.${MADLIB_VERSION_MINOR}.${MADLIB_VERSION_PATCH}")
810
set(MADLIB_PGXN_NAME "madlib-pgxn-${MADLIB_PGXN_VERSION_STR}")
911

1012
configure_file(META.json.in META.json)

‎deploy/gppkg/CMakeLists.txt‎

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -2,7 +2,7 @@
22
# Packaging for Greenplum's gppkg
33
# ------------------------------------------------------------------------------
44

5-
set(MADLIB_GPPKG_VERSION "1.7")
5+
set(MADLIB_GPPKG_VERSION "1.7.1")
66
set(MADLIB_GPPKG_RELEASE_NUMBER 1)
77
set(MADLIB_GPPKG_RPM_SOURCE_DIR
88
"${CMAKE_BINARY_DIR}/_CPack_Packages/Linux/RPM/${CPACK_PACKAGE_FILE_NAME}"

‎deploy/postflight.sh‎

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -2,7 +2,7 @@
22

33
# $0 - Script Path, $1 - Package Path, $2 - Target Location, and $3 - Target Volumn
44

5-
MADLIB_VERSION=1.2
5+
MADLIB_VERSION=1.3
66

77
find /usr/local/madlib/bin -type d -exec cp -RPf {} /usr/local/madlib/old_bin \; 2>/dev/null
88
find /usr/local/madlib/bin -depth -type d -exec rm -r {} \; 2>/dev/null

‎src/config/Version.yml‎

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -1 +1 @@
1-
version: 1.2
1+
version: 1.3

‎src/madpack/changelist.yaml‎

Lines changed: 30 additions & 4 deletions
Original file line numberDiff line numberDiff line change
@@ -1,4 +1,4 @@
1-
# Changelist for MADlib version 1.1 to 1.2
1+
# Changelist for MADlib version 1.2 to 1.3
22

33
# This file contains all changes that were introduced in a new version of
44
# MADlib. This changelist is used by the upgrade script to detect what objects
@@ -9,11 +9,11 @@
99
# file installed on the upgrade version. All other files (that don't have
1010
# updates), are cleaned up to remove object replacements
1111
new module:
12-
arima:
13-
arima_forecast:
1412

1513
# Changes in the types (UDT) including removal and modification
1614
udt:
15+
__logregr_result:
16+
linregr_result:
1717

1818
# List of the UDF changes that affect the user externally. This includes change
1919
# in function name, change in argument order or argument types, and removal of
@@ -22,10 +22,36 @@ udt:
2222
# are user views dependent on this function, since the original function will
2323
# not be present in the upgraded version.
2424
udf:
25+
# linear regression: 'num_processed' added in 'linregr_result'
26+
- linregr_final:
27+
rettype: schema_madlib.linregr_result
28+
argument: schema_madlib.bytea8
29+
- linregr_merge_states:
30+
rettype: schema_madlib.bytea8
31+
argument: schema_madlib.bytea8, schema_madlib.bytea8
32+
- linregr_transition:
33+
rettype: schema_madlib.bytea8
34+
argument: schema_madlib.bytea8, double precision, double precision[]
35+
36+
# logistic regression: 'num_processed' added in '__logregr_result'
37+
- __logregr_cg_result:
38+
rettype: schema_madlib.__logregr_result
39+
argument: double precision[]
40+
41+
- __logregr_irls_result:
42+
rettype: schema_madlib.__logregr_result
43+
argument: double precision[]
44+
45+
- __logregr_igd_result:
46+
rettype: schema_madlib.__logregr_result
47+
argument: double precision[]
2548

2649
# Changes to aggregates (UDA) including removal and modification
2750
# Overloaded functions should be mentioned separately
2851
uda:
29-
52+
- linregr:
53+
rettype: schema_madlib.linregr_result
54+
argument: double precision, double precision[]
55+
3056
# Cast operators (UDC) updated/added in v1.1
3157
udc:

src/madpack/changelist_1.0_1.2.yaml renamed to src/madpack/changelist_1.0_1.3.yaml

Lines changed: 32 additions & 4 deletions
Original file line numberDiff line numberDiff line change
@@ -1,4 +1,4 @@
1-
# Changelist for MADlib version 1.0 to 1.2
1+
# Changelist for MADlib version 1.0 to 1.3
22

33
# This file contains all changes that were introduced in a new version of
44
# MADlib. This changelist is used by the upgrade script to detect what objects
@@ -17,7 +17,8 @@ new module:
1717

1818
# Changes in the types (UDT) including removal and modification
1919
udt:
20-
20+
__logregr_result:
21+
linregr_result:
2122
# List of the UDF changes that affect the user externally. This includes change
2223
# in function name, change in argument order or argument types, and removal of
2324
# the function. In each case, the original function is as good as removed and a
@@ -30,9 +31,36 @@ udf:
3031
rettype: schema_madlib.matrix_result
3132
argument: matrix_in text, matrix_out text
3233

34+
# linear regression: 'num_processed' added in 'linregr_result'
35+
- linregr_final:
36+
rettype: schema_madlib.linregr_result
37+
argument: schema_madlib.bytea8
38+
- linregr_merge_states:
39+
rettype: schema_madlib.bytea8
40+
argument: schema_madlib.bytea8, schema_madlib.bytea8
41+
- linregr_transition:
42+
rettype: schema_madlib.bytea8
43+
argument: schema_madlib.bytea8, double precision, double precision[]
44+
45+
# logistic regression: 'num_processed' added in '__logregr_result'
46+
- __logregr_cg_result:
47+
rettype: schema_madlib.__logregr_result
48+
argument: double precision[]
49+
50+
- __logregr_irls_result:
51+
rettype: schema_madlib.__logregr_result
52+
argument: double precision[]
53+
54+
- __logregr_igd_result:
55+
rettype: schema_madlib.__logregr_result
56+
argument: double precision[]
57+
3358
# Changes to aggregates (UDA) including removal and modification
3459
# Overloaded functions should be mentioned separately
3560
uda:
36-
61+
- linregr:
62+
rettype: schema_madlib.linregr_result
63+
argument: double precision, double precision[]
64+
3765
# Cast operators (UDC) updated/added in v1.1/v1.2
38-
udc:
66+
udc:

‎src/madpack/changelist_1.1_1.3.yaml‎

Lines changed: 61 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,61 @@
1+
2+
# Changelist for MADlib version 1.1 to 1.3
3+
4+
# This file contains all changes that were introduced in a new version of
5+
# MADlib. This changelist is used by the upgrade script to detect what objects
6+
# should be upgraded (while retaining all other objects from the previous version)
7+
8+
# New modules (actually .sql_in files) added in upgrade version
9+
# For these files the sql_in code is retained as is with the functions in the
10+
# file installed on the upgrade version. All other files (that don't have
11+
# updates), are cleaned up to remove object replacements
12+
new module:
13+
arima:
14+
arima_forecast:
15+
16+
17+
# Changes in the types (UDT) including removal and modification
18+
udt:
19+
__logregr_result:
20+
linregr_result:
21+
22+
# List of the UDF changes that affect the user externally. This includes change
23+
# in function name, change in argument order or argument types, and removal of
24+
# the function. In each case, the original function is as good as removed and a
25+
# new function is created. In such cases, we should abort the upgrade if there
26+
# are user views dependent on this function, since the original function will
27+
# not be present in the upgraded version.
28+
udf:
29+
# linear regression: 'num_processed' added in 'linregr_result'
30+
- linregr_final:
31+
rettype: schema_madlib.linregr_result
32+
argument: schema_madlib.bytea8
33+
- linregr_merge_states:
34+
rettype: schema_madlib.bytea8
35+
argument: schema_madlib.bytea8, schema_madlib.bytea8
36+
- linregr_transition:
37+
rettype: schema_madlib.bytea8
38+
argument: schema_madlib.bytea8, double precision, double precision[]
39+
40+
# logistic regression: 'num_processed' added in '__logregr_result'
41+
- __logregr_cg_result:
42+
rettype: schema_madlib.__logregr_result
43+
argument: double precision[]
44+
45+
- __logregr_irls_result:
46+
rettype: schema_madlib.__logregr_result
47+
argument: double precision[]
48+
49+
- __logregr_igd_result:
50+
rettype: schema_madlib.__logregr_result
51+
argument: double precision[]
52+
53+
# Changes to aggregates (UDA) including removal and modification
54+
# Overloaded functions should be mentioned separately
55+
uda:
56+
- linregr:
57+
rettype: schema_madlib.linregr_result
58+
argument: double precision, double precision[]
59+
60+
# Cast operators (UDC) updated/added in v1.1
61+
udc:

0 commit comments

Comments
 (0)