Notice: This Wiki is now read only and edits are no longer possible. Please see: https://gitlab.eclipse.org/eclipsefdn/helpdesk/-/wikis/Wiki-shutdown-plan for the plan.
OSEE/ReqAndDesign
Contents
Fast, highly relevant search
Requirements
- support regular expression searches
- support case-insensitive and case-sensitive (use second pass)
- history on name and url changes
- history on search terms changes
- access control
Design
- Query Parser Steps
- make search text lowercase
- split search text into search terms by white space, allow quotes to enclose a search term that includes white space
- strip leading and trailing characters that are neither letters or digits
- remove trailing 's (possessive case)
- discard single char terms that are not digits (only implement if proven useful)
- transform words into singular form
- discard duplicate terms
- discard stop words. see http://www.gobloggingtips.com/wp-content/uploads/2014/08/Google-stopwords.txt
- truncate term if longer than 128 characters (max token length)
- sort terms
- compute query_hash with Arrays.deepHashCode() on the sorted terms from previous step
source text | options | matching query | non-matching query |
---|---|---|---|
C. S. Lewis | tbd | C S Lewis | CS Lewis |
75 | tbd | 7 | 0.7 |
75% | tbd | 75 | % |
nav_power_on | tbd | power_on | power on |
Books | tbd | booK | bookes |
source text | tbd | matching query | non-matching query |
- Query options
-
- Case sensitive
- Match word order (no effect with 1 search term)
- Include leading punctuation
- Include trailing punctuation
- Whole content match
- Whole word match
- Regular expression match- a list of regular Expressions that are individually matched against terms in the first pass, and a single regular expression that is applied in the second pass or blank to skip second pass (no other option should be combined with this one)
Data Model
osee_search_url(BIGINT url_id, Varchar(128) title, Varchar(2048) url, valid_test_time) PRIMARY KEY (url_id) ORGANIZATION INDEX
- title - search terms will be extracted from the title using the query parser steps and stored with rank name
- url uniqueness is enforced via atomic insert command
- INSERT INTO osee_search_url m1 (match_id, name, url, valid_test_time) SELECT ?,?,?,? FROM dual WHERE NOT EXISTS (SELECT 1 From osee_search_url m2 WHERE m2.url = ?)
- maintenance operation identifies dead links. valid_test_time is updated with current date and time when the link is tested as valid
osee_search_term(Varchar(128) text, SMALLINT match_type_id, BIGINT app_id, SMALLINT rank_type_id, BIGINT term_group_id, SMALLINT term_group_size, BIGINT match_id) PRIMARY KEY (text, match_type, term_group_id, match_id) ORGANIZATION INDEX COMPRESS 2
- text is normalized search term (see query parser steps)
- match type
- artifact, attribute, relation (for rationale), url type
- app_id
- match_id can be a osee_search_url.url_id or an attribute or relation gamma_id (or possibly extended to any other id)
- the rank allows early exclusion of unwanted term types such as content with the sql criteria that only includes rank > 20
- Results are returned in ascending rank order of the sum of all ranks for a match_id
- rank_type_id:
- Tuple3Type<RankTypeId, Long, String> RankType = Tuple3Type.valueOf(SearchFamily, 12L); // rank_type, rank (larger is higher), rank
- 1 [220] unique id - (employee id, charge number, USPS Tracking number, etc.)
- 2 [200] keyword - terms that when searched together should represent a top result
- 3 [ 40] tag (meta) - terms applied by human intelligence to classify and categorize
- 4 [ 20] content - terms extracted from the content body using the query parser steps
- 5 [ 10] team - relevant to a given team
type name
- term_group_id - 1 when the term_group_size is 1; otherwise randomly generated id used to group search terms to apply a rank only when all terms in the group match
osee_search_query(BIGINT query_hash, Varchar(500) query_text, SMALLINT status, BIGINT top_match)
- CREATE INDEX OSEE_SEARCH_QUERY_TEXT_IDX ON OSEE_SEARCH_QUERY (QUERY_TEXT);
- query_hash is computed using the Query Parser Steps
- query_text exactly as entered
- status
- no results
- unverified results
- verified results
- top_match is the top match returned the last time this query was run. Useful for sharing links/ searches. consider supporting more than 1 top matches
- search prediction based on query_text with past queries of user sorted first
osee_search_share(BIGINT to_user, BIGINT from_user, BIGINT query_hash, timestamp time) PRIMARY KEY (to_user, from_user, query_hash) ORGANIZATION INDEX COMPRESS 2
osee_search_history(BIGINT item_id, BIGINT user_id, timestamp) PRIMARY KEY (item_id, user_id, timestamp) ORGANIZATION INDEX COMPRESS 2
Rest Interface
- Allow request to specify max number of links or all (default is 25). Top ranked n links are returned as json {id, name, url, search terms, rank}
- in order to set access control on a link you have to have the access to the group you are setting it to, you can add anyone else
User Interface
When entering a new URL the name is initially populated from the web page title
Search results will include a why link that provides the rank information and allows users to improve this information
Fast, versioned tuple service
Requirements
- Creation and deletion of tuples shall have transactional history
- tuple operations shall be near constant time
- shall efficiently support tuples of numbers and strings of length 2, 3, and 4
Uses Cases
- multiple tags per transaction with with history
osee_tuple3(tx_key_type, branch_id, tx_id, attr_id, gamma_id)
- cross branch linking
osee_tuple2(relation_type, a_art_id, b_art_id, gamma_id)
- user defined tables
osee_tuple2(BIGINT my_table_type, BIGINT row_id, String row_data, gamma_id)
- the row_id allows the rows to be returned in order, also allows showing history of a given row
- the row_data stores the json representing the row data
Design
Data Model
- Tuples provide a high performance general purpose storage that leverages index-organized tables. Traditional database tables are unorganized heaps that require indexes for fast access. If the size of individual rows can be kept small, then index-organized tables can be used and remove the overhead and duplication of also having the unorganized heap. Additionally, these tables support key compression which further improves performance and reduces storage requirements.
osee_tuple2 (tuple_type, element1, element2, gamma_id) PRIMARY KEY (tuple_type, element1, element2) ORGANIZATION INDEX COMPRESS 2; osee_tuple3 (tuple_type, element1, element2, element3, gamma_id) PRIMARY KEY (tuple_type, element1, element2, element3) ORGANIZATION INDEX COMPRESS 2; osee_tuple4 (tuple_type, element1, element2, element3, element4, gamma_id) PRIMARY KEY (tuple_type, element1, element2, element3, element4) ORGANIZATION INDEX COMPRESS 3; unique index: gamma_id
- Each tuple has a type. See org.eclipse.osee.framework.core.enums.CoreTupleTypes
- gamma_id provides for branching and history. Tuples can be created and deleted on a branch using transactions
- elements can be of type number or string. A string is stored as a key to the osee_key_value table
- In the table each tuple element is stored as a BIGINT and if it represents a String then this number is a key into the OSEE_KEY_VALUE table where the string is stored. This approach keeps each row in a tuple table to be of a small fixed size.
Product Line Engineering
Requirements
- User can quickly select branch-view without requiring knowledge of corresponding version of product line. Select by branch-view name (i.e. Country X San) which is short hand for selecting My Product Line 2.0 branch with Country X view.
- View applicability has history and can vary independently by branch.
- External tools can join on view applicability in their sql queries
- Need a list of branch-views to show in branch selection dialogs
Definitions
- Applicability_Id - just an id
- part of osee_txs table, and thus versioned.
- applicability id per artifact, attribute and relation.
- applicability can be on sub attribute level using grammar within Word or other content
- Feature - decomposed item of product or configuration
- is an Applicability_Id
- represented as unique NAME, ID
- can be a single Feature or some mathmatical combination of Features. eg: IDM, ARC210, IDM & ARC210
- Variant - combination of Features that make up a version of a product
- is an Applicability_Id
- represented as unique NAME, ID
- View - cross-section of a Branch that shows items applicable to that view
- either a Configuration, Feature or grouping
- read-only. Editing can only be done on parent Branch
- what is shown is computed from selected Configs, Features
Design
Data Model
- item applicability is fully transactional on all branches
osee_txs (branch_id, transaction_id, gamma_id, tx_current, mod_type, app_id)
Default value for app_id in osee_txs is 1 which means in all views (base). View artifacts are on the product line branches.
- view applicability
osee_tuple2 (CoreTupleTypes.ViewApplicability, ArtifactId view, String applicabilityText, gammaId)
The view applicability tuples and view artifacts are on the product line branches thus maintaining full transactional history. The applicability text is the feature applicability text or the name of the of the view for non product line engineering use of views. Each view will be mapped to applicability id 1 (Base) so that base will be included for all configs by simply joining on the view_id. The featureApplicabilityId is the randomly generated key to applicabilityText in osee_key_value.
- named branch-views
osee_tuple3 (CoreTupleTypes.BranchView, BranchId branch, ArtifactId view, String name, BIGINT gammaId)
Views are stored as artifacts of type "Branch View" on the branch their are views of. Named branch-views are stored as tuples on the common branch. These are used to populate the list of branches faster than searching and loading every "Branch View" on every branch. The branch view is accessed via BranchId.getView()
- feature applicability definition
osee_tuple4 (app_feature_tuple_type, app_id, feature_id, value_attr_id, and_or_flag, gamma_id)
The Feature artifact has an attribute for each possible value. It also has a name, description, abbreviation, and type (single or multi-valued).
osee_tuple3 (compound_app_tuple_type, app_id, app_id, and_or_flag, gamma_id)
- feature applicability involving more than one feature is stored as a list of feature applicability (recursive to any depth).
-- return all items (referenced by gamma_id) currently in a view of a branch SELECT * FROM osee_txs txs, osee_tuple2 app WHERE branch_id = ? AND tx_current = 1 AND txs.app_id = app.e2 AND e1 = ?; -- applicability clause in SQL is only applied when using a branch view -- select applicability for a given view (e1) SELECT e2, VALUE FROM osee_txs txs, osee_tuple2 app, osee_key_value WHERE tuple_type = 2 AND e1 = ? AND app.gamma_id = txs.gamma_id AND branch_id = ? AND tx_current = 1 AND e2 = KEY;
Activity Logging and Monitoring
Requirements
- shall handle creation/update of fine-grained log entries for at least 500 concurrent users
- shall support logging by OSEE and other applications
- the web of log entries related to an individual instance of a user request shall be able to be hierarcically related
- log entries shall be quickly accessible based on any combination of source, user, timestamp, log type, duration, status
- log entries shall be accessible (especially) when an application server is unresponsive
- log entries shall be available until they are deleted by an admin or admin policy (applied by server automatically)
- at run-time logging shall be enabled/disabled based on any combination of user, source, log level, and type
- access control shall be applied at the log entry type basis
Design
Data Model
- osee_activity db table
- Log entry in Java: BIGINT entryId, BIGINT parentId, BIGINT typeId, BIGINT startTime, BIGINT duration, BIGINT agentId, BIGINT status, String msgArgs
- entry_id - random long returned for log method call
- parent_id - id of entry used for grouping this and related entries. For root entries, it is the negative of session id of the client or the server id. Ranges are used to group by client/server kind (IDE client, app server, rest client).
- type_id - foreign key to type_id in osee_log_type table
- start_time - long with ms since epoch
- duration - starts at -1 and is never updated if duration does not apply, otherwise updates when the associated job ends with duration in ms
- account_id - long account id (the account_id returned from account management services
- status:
0 initial value 1-99 percent complete 100 completed normally 101 completed abnormally
- msg_args newline separated list of strings used with String.format(msg_format, msg_args);
- Each new log entry's parent_id, agent_id is mapped to the thread that created it (only the most recent mapping per thread is maintained)
- When an exception is thrown, it is logged as a child of the parent corresponding to the current thread. If no mapping is found in ConcurrentHashMap<Thread, Pair<Long, Long>>()
- Log entry type in DB: type_id, log_level, software_unit, message_format
- type_id - a fine-grained application defined type, random id, defined as tokens and stored in the db for cross application support
- log_level - as defined by java.util.logging.Level
- module - application defined name of the software unit that uses this log entry type
- msg_format - format defined by java.util.Formatter or if blank the raw message details are used directly
High Performance
- 2 ConcurrentHashMap are allocated with an initial configurable size: newLogEntires, updatedEntries
- newly created log entries are added to newLogEntires using the entry_id as the key and the array of sql insert parameters as the value
- updated log entries are checked for in newLogEntires and updated if they exist, otherwise the update map is checked and updated if exists, else added to updatedEntries
- A timer tasks runs at a configurable (short) periodic rate and batch inserts the log entries in the insert map and then runs the updates. This means that any update to a log entry that occurs in less than this configured time will not require a database update (i.e. writing the duration of a short operation). This also means only one thread writes to the log table per JVM.
- new DrainingIterator(newLogEntires.values().iterator()) is used to iterate through the values and remove them one at a time during the batch insert
- upon server shutdown must flush log
- IDE client will directly use the same service that is used on the server
- data structure options
- Optimize JDBC Performance
Java API
Long createThreadEntry(long userId, Long typeId); Long createThreadEntry(long userId, Long typeId, long parentId); Long createEntry(Long typeId, Object... messageArgs); Long createEntry(Long typeId, Long parentId, Object... messageArgs); void updateEntry(Long entryId, Long status); Long createExceptionEntry(Throwable throwable);
- The first interface to the logging data can be the basic REST navigation
Exception Handling
Requirements
- avoid unnecessary wrapping of exceptions
Design
Checked exceptions I love you, but you have to go Why should you use Unchecked exceptions over Checked exceptions Clean Code by Example: Checked versus unchecked exceptions
- Use application specific exceptions that extend RuntimeException - application specific allows for setting exception breakpoints in the debugger
- Do not declare any run-time exceptions in any method signatures
Artifact Query and Loading
Requirements
- As a developer and user, I would like quick searching/loading in as few sql statements provided from the server in a single call
- Allow search rest call on server to load artifacts and just send data back necessary to construct artifacts/relations instead of client making more database calls to load
- Consolidate the searching/loading from 8 queries to 1 or 2
- As a developer, I would like to specify which relations and attributes and artifact data to load for the results of my query.
- Provide selective loading
- As a developer, I would like to search for artifacts and then load related artifact to a certain level. Example: ATS loads 3 Team Workflows, then needs to bulk load related actions, tasks and reviews.
- Add ability to bulk load to search criteria. Include load level and list of relations to follow.
- As a developer, I would like to be able to load artifacts with only artifact data (ids, types, etc) and a few attributes like Name. Example: Artifact Explorer shouldn't need to load the entire artifact when only the Name, Type and whether it has children is needed. Another example is Quick Search results, where all you need is the basic data from the results until the user is interested in selecting a resulting item.
- Allow ability to specify artifact data and minimal attrs or relations.
- Also provide case where want to "Show Relations" which would require knowing what relations exist and their types.
- fast, scalable
Design
Dynamic Web App
Requirements
- As a developer and user, I would like to be able to specify the content of a web page by providing a JSON description.
- Allow a group of controls to be displayed, and to specify REST calls for obtaining content and changing content.
Design
We have adopted JSON Forms to supply this capability, but have written our own Angular JS controls to support the capability to have each control have its own REST calls. All of the controls in the box below additionally provide links for getting and setting data. This allows the page to update and check data one control at a time.
Text box control: Similar to the JSON Forms text control. Checkbox control: Similar to the JSON Forms checkbox. Multi-line text box control: Provides a resizeable multi line text box. Dropdown control: Similar to the JSON Forms dropdown control. Multi-select Dropdown control: Wraps popular dotansimha Angular JS Multiselect Dropdown. Program-version control: a special case of the multi select dropdown to support selection of OSEE ATS program and version. Table control: wraps Angular UiGrid control to allow for table data.
Distributed OSEE Repositories
Level 0 capability
Branches are private to their own OSEE repository, but distributed IDs are used to avoid future ID collisions between OSEE repositories. Implementation Note: IDs are randomly generated, except transactions which are numbered sequentially on a per branch basis (avoids transaction collisions since transaction ids can't be random due to max transaction calculations)
Level 1 capability
A branch may be cloned from the source OSEE repository into a destination repository as a read-only branch that will automatically update when network connectivity exists between the two repositories. Note: any repository may request to clone a branch from another repository. If the request is granted the source repository marks its branch as source and then allows the clone.
Level 2 capability
The branch updates for a cloned branch will also be able to be performed via an export of the source updates to the file system followed by an import in the OSEE that has the cloned branch.
Level 3 capability
A branch may be bidirectionally shared between OSEE repositories. The OSEE repository of the source branch tracks the branch lock. Other OSEE repositories start with a clone of the shared branch. In order for any OSEE to commit a transaction to the shared branch, it will first acquire the branch lock from the source repository. All transactions from the source are copied to the other repositories.
Fast, Robust Attribute Storage
Requirements
- System can store and provide fast access to small attributes in bulk
- System can store and provide fast access at least 1 MB attributes
- System can store at least 1GB attributes
Design
Data Model
- each osee attribute type explicitly lists supported search types: exact match, case insensitive exact match, partial match, or none
osee_search_hash ( match_type app_id term_hash art_id gamma_id)
- osee_attribute.value stores boolean, numeric, enumeration, date, strings, and binary
osee_attribute (art_id, type_id, item_id, order, gamma_id, value)
Binary data is rare and often large and is never searched directly (instead it is tokenized and indexed) so Store it in an iot with the gamma_id as key
-- return all attributes for a given artifact currently in a view of a branch SELECT * FROM osee_txs txs, osee_tuple2 app WHERE branch_id = ? AND tx_current = 1 AND txs.app_id = app.e2 AND e1 = ?; -- applicability clause in SQL is only applied when using a branch view -- select applicability for a given view (e1) SELECT e2, VALUE FROM osee_txs txs, osee_tuple2 app, osee_key_value WHERE tuple_type = 2 AND e1 = ? AND app.gamma_id = txs.gamma_id AND branch_id = ? AND tx_current = 1 AND e2 = KEY;
REST
Design
- The path component contains hierarchical data that, along with data in the non-hierarchical query component serves to identify a resource RFC3986 3.3 first paragraph
- name of JSON keys should be camelCase just like Java
- each segment in the REST URL has a singular name even the name of a collection of resources
- GET/HEAD/PUT/DELETE shall not use a request message body according to originator of REST Roy T. Fielding
- URL all lowercase with words separated with hyphens
- MUST use POST when the action is not idempotent. GET, PUT and DELETE methods are required to be idempotent.
- REST APIs must be hypertext-driven