Pith. sign in

REVIEW 4 major objections 6 minor 102 references

CODENS claims that pull-request history can be turned into a typed, queryable knowledge graph that yields highly relevant and well-grounded answers to repository-level developer questions.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

T0 review · deepseek-v4-flash

2026-08-01 17:01 UTC pith:ONC27CM5

load-bearing objection CODENS is a well-written integration of PR replay and typed knowledge graphs, but its own pipeline description drops new files added after the initial scan — a real gap under the 'living documentation' claim. the 4 major comments →

arxiv 2607.18356 v1 pith:ONC27CM5 submitted 2026-07-20 cs.SE cs.AI

CODENS: Transforming Code Changes into Living, Accessible, and Queryable Documentation

Abdelhak Kelious , Chyrine Tahri , Eliot Bardet This is my paper
classification cs.SE cs.AI
keywords knowledge graphcode documentationretrieval-augmented generationpull request analysisagentic retrievalsoftware maintenanceliving documentationmodel-view-controller frameworks
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

This paper claims that software documentation can be kept alive automatically by converting pull requests into structured knowledge. It describes CODENS, a pipeline that builds a typed graph from a model-view-controller codebase's architectural conventions, then replays pull requests to enrich each component with semantic attributes and typed relations. The resulting graph is queried through three retrieval modes, the deepest being an agent that traverses the graph with purpose-built tools. On an 11-question industrial evaluation, answers scored well on relevance, completeness, and grounding, though reviewers found them too verbose and code-centric for documentation use.

Core claim

The central claim is that a schema-driven, state-injecting extraction pipeline can turn pull-request diffs into cumulative component knowledge. For each changed file, CODENS feeds a large language model the existing node state plus the new diff, so the model updates rather than replaces prior semantics. After processing pull requests chronologically, an edge extractor resolves typed relations such as invokes-model, renders-view, and tests-component to form a graph with hundreds of unique edges over thousands of skeleton nodes. The paper reports perfect context precision and faithfulness across 11 production questions, with an average answer relevancy of 0.94 and human relevance scores around

What carries the argument

The engine is the state-injection merge loop: the current semantic representation of a graph node is injected alongside each new pull-request diff into an LLM, which returns schema-valid JSON attributes that are merged (scalar fields replaced, lists unioned, provenance tracked). This loop lets incremental PR processing accumulate into a typed software knowledge graph whose node embeddings support three retrieval modes: vector search, automatic multi-hop expansion, and agent-guided traversal through tools that read nodes, list neighbors, and issue graph queries.

Load-bearing premise

The system's reliability rests on the LLM faithfully merging prior node state with each new diff into schema-valid, semantically accurate JSON, with no verification loop to catch or correct a bad extraction.

What would settle it

Replay a controlled sequence of pull requests affecting a single component, with the final source code fixed; after each PR, check whether the node's accumulated description actually matches the current code. A single bad extraction that persists into later states and silently contradicts the source would falsify the claim that the graph tracks the repository.

Watch this falsifier. Get emailed when new claim-graph text bears on it.

Share X Bluesky LinkedIn Reddit HN

If this is right

  • If CODENS works as described, merging a pull request automatically refreshes the documentation for every component it touches, removing the separate documentation-writing step.
  • Developers can ask natural-language questions about features, flows, and business rules and receive answers whose evidence is traceable to graph nodes and typed relations.
  • Each node preserves a semantic change history, giving a chronological rationale for why a component evolved the way it did.
  • The three retrieval modes form a cost-completeness tradeoff: agent-guided traversal produces the most complete answers at the highest token and time cost.
  • The evaluation suggests that the remaining bottleneck is answer presentation, not retrieval or grounding.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The state-injection design implies that documentation quality degrades gracefully only if extraction errors are rare; a natural testable extension is a confidence gate that flags low-confidence LLM updates for human review.
  • The framework-convention schema might transfer to other MVC-like codebases, making PR-driven documentation a general pattern rather than a Rails-only tool.
  • The identified verbosity problem suggests a two-stage synthesis: first retrieve graph evidence, then compress it into user-facing abstractions such as page paths and UI flows.
  • If pull requests themselves become documentation events, code review could double as documentation review, changing how teams budget for maintenance.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

4 major / 6 minor

Summary. The paper describes CODENS, a system that mines a Ruby on Rails repository and its pull-request history to incrementally build a typed knowledge graph of software components. CODENS first creates skeleton nodes for framework component types (17 types, 1,739 nodes in the reported project), then replays PRs chronologically: for each changed file it injects the current node state and the diff into GPT-4, extracts schema-defined semantic attributes, merges them while preserving per-PR provenance, and later derives typed edges (622 edges, 11 relationship types). The graph is stored in Neo4j with vector embeddings and exposed through three retrieval modes: standard vector retrieval, graph-expanded multi-hop retrieval, and a ReAct agent with graph tools. The evaluation on one production codebase with 11 anonymized questions reports human scores averaging 4.09/5 relevance, 4.45/5 completeness, and 4.91/5 document relevance, perfect automatic context precision and faithfulness, and qualitative feedback asking for more concise, user-oriented synthesis.

Significance. The work addresses a real and well-motivated problem: PR-level design knowledge is typically lost unless someone writes documentation. The state-injection design, in which the LLM is prompted with the existing node state and the current diff so that semantic updates accumulate rather than overwrite, is a sensible and potentially useful idea. The provenance/change-history tracking and the operational metrics panel (latency, tokens, cost, carbon) are thoughtful additions for industrial deployment. The graph-agent retrieval design is a plausible extension of RAG for code artifacts. However, the evidence presented is currently too thin to substantiate the abstract's claim that CODENS 'produces highly relevant and well-grounded answers': only 11 questions from a single codebase, a single human evaluator, no baseline comparison, and no statistical testing. The paper reads as an exploratory case study or system demonstration, not as a validated research result.

major comments (4)
  1. [§3, Tables 2–3] The evaluation does not support the abstract's claim that 'CODENS produces highly relevant and well-grounded answers.' The evidence is 11 anonymized questions from a single codebase, with a single human evaluator (the project lead). There is no baseline comparison (e.g., standard vector RAG, GraphRAG, RepoCoder, or a non-graph RAG), no statistical testing or error bars, and only the agent retrieval mode is evaluated—the standard and multi-hop modes are never assessed. The paper itself calls the study 'exploratory,' but §3's 'These results indicate...' and the abstract overstate the findings. At minimum, the authors should add a baseline and report per-mode results, or explicitly reframe the claims as a case study with appropriate caveats.
  2. [§2.2, Step 1] The pipeline has a gap for changed files that do not correspond to skeleton nodes. The initial scan in §2.1 is described as scanning 'the repository' and creating 1,739 nodes, while Step 1 then 'retain[s] only files matching existing nodes.' If the scan is of the current repository snapshot, files added by later PRs that still exist are covered; if the scan is of an earlier historical state, any file added during the PR range is silently dropped. In either reading, files that were added and later deleted, or renamed away from a path that no longer exists, can never receive a node or enter the graph. Since the central value proposition is that PR history becomes 'living documentation,' the authors must either create nodes for newly appearing paths during replay or explicitly scope the system to the current file inventory and report how many added/deleted/renamed files occurred in the repl
  3. [§2.2, Steps 2–4] The pipeline assumes GPT-4 will produce schema-valid, semantically accurate node updates from the diff-plus-state prompt, and then applies them automatically: scalar fields are replaced (with the old value archived), list fields are merged by uniqueness. There is no verification loop, confidence threshold, or mechanism to detect and correct a bad extraction. Because the updated state is fed back as the 'prior state' for all subsequent PRs, a single erroneous extraction can propagate through the graph and degrade later QA answers. The manuscript should include at least a manual error analysis of a sample of merged node states, or add a validation/rollback step, before the extraction pipeline can be considered reliable.
  4. [§3, automatic metrics] The reported perfect context precision and faithfulness (1.00 for all 11 queries) are not informative evidence of retrieval quality as described in the paper. If the LLM is instructed to answer from the retrieved context, faithfulness to that context is almost guaranteed by construction, and context precision computed on the system's own retrieved set is not a meaningful quality measure. No definitions or computation details are given for these metrics (e.g., which model or human computes them, what the reference set is). The authors should either define the metrics and include a baseline, or withdraw them and rely on direct human assessment of the final answer.
minor comments (6)
  1. [§2.4] First sentence: '...with three retrieval modes of increasing depth, The chatbot' — comma splice; should be a period and capital 'T'.
  2. [§2.2, Step 1] Clarify what 'existing nodes' means after the initial scan—current repository snapshot or a historical checkout. This is central to interpreting the replay and should be stated explicitly.
  3. [§2.2–2.4] Report the values of the free parameters: top_k, multi-hop traversal depth / neighbor filter threshold, the full 17-type component taxonomy, the cue-based edge regex taxonomy, and the exact prompt templates. Without these, the reported scores cannot be reproduced or interpreted with respect to parameter sensitivity.
  4. [Table 2] The 'Avg' column is undefined. State clearly whether it is the mean of the three human scores, a composite automatic score, or something else. Also clarify how the automatic 'Ans.' (answer relevancy) metric is computed.
  5. [§2.2, Step 4] The phrase 'semantic change history' is not formally defined. The text suggests that scalar-field history is archived per PR, but list fields are merged without per-element provenance; state this explicitly.
  6. [§4] The conclusion suggests 'possible operational advantages' in cost, token usage, and response time, but no comparison to any alternative system is made. This sentence should be removed or supported by a baseline comparison.

Circularity Check

0 steps flagged

No significant circularity: CODENS is an implemented data-processing pipeline; its evaluation limitations are validity and generality concerns, not circular reductions.

full rationale

CODENS's claimed derivation is a sequential pipeline: repository scan creates skeleton nodes, PR diffs are injected with current node state into an LLM, schema-validated JSON updates are merged into a graph, edges are extracted, and retrieval modes query the resulting Neo4j store. None of these steps defines an output in terms of the claim it is used to support. The evaluation mixes human ratings (relevance, completeness, document relevance) with automatic RAG metrics; while the automatic metrics are computed on the system's own retrieved contexts and could be influenced by prompt design, the paper does not define either the metric or the answer generation so that a high score is guaranteed by construction. The authors cite only external prior work (KG4Py, CodexGraph, RepoCoder, GraphCoder, GraphRAG, ReAct, SWE-agent), with no load-bearing self-citation chain. The structural gap in §2.2 step 1 — retaining only files matching existing nodes, so new files in PRs are never added — is a correctness and completeness threat to the 'living documentation' claim, but it is not circularity: the system's outputs are not equivalent to its inputs. The reported limitations (single codebase, 11 questions, need for broader comparative benchmark) are evaluation-scope issues, not evidence that a prediction reduces to a fit or to a self-citation.

Axiom & Free-Parameter Ledger

5 free parameters · 6 axioms · 0 invented entities

CODENS introduces no new physical or mathematical entities. Its reliance is on engineering design choices—hand-authored schemas, prompts, thresholds, and embeddings—plus an implicit trust in LLM extraction and a single-project evaluation. These assumptions, not fitted constants, are what a replication attempt would need to test.

free parameters (5)
  • top_k seed retrieval count = 3
    Hand-picked; the agent mode starts from the top-3 vector matches, with no ablation or sensitivity analysis.
  • multi-hop traversal depth / neighbor filter threshold = 1–2 hops; threshold unspecified
    Design choices that directly affect retrieval quality; no experiments vary them.
  • component type taxonomy size = 17 component types
    Hand-defined schema based on Rails directories; all downstream extraction depends on this taxonomy.
  • cue-based edge regex taxonomy = regex patterns from a taxonomy file
    Hand-authored patterns used for cue-based edge extraction; no evaluation of coverage or precision.
  • merge instruction prompt design = GPT-4 prompt with state injection
    The central behavior of updating existing node state instead of replacing it is prompt-driven; the prompt is not public and no ablation is reported.
axioms (6)
  • domain assumption Rails directory conventions deterministically classify each file's architectural role.
    Initialization creates skeleton nodes from directory-based component types (§2.1). If conventions are violated or files are nested unusually, the graph is mislabeled from the start.
  • domain assumption GPT-4 reliably produces schema-valid, faithful semantic extractions from diff plus state.
    The entire graph enrichment in §2.2 step (3) depends on LLM extraction accuracy. There is no described validation or correction mechanism for bad extractions.
  • domain assumption State injection causes the LLM to update rather than overwrite prior knowledge.
    Section 2.2 step (2) assumes that feeding previous state plus diff yields integrated representations. If the LLM silently drops old facts, history is lost despite the merge strategy.
  • domain assumption all-mpnet-base-v2 embeddings capture semantic similarity between natural-language questions and code documentation text.
    Vector retrieval in §2.3 uses cosine similarity over these embeddings; no evaluation compares embedding choices or measures retrieval recall independently.
  • domain assumption GitHub API diff replay captures all meaningful design knowledge.
    The system processes only changed files matching existing nodes (§2.2). Architectural decisions not visible in diffs, or entirely new files not yet matched, are missed.
  • domain assumption The lead developer's 1–5 ratings are reliable ground truth for answer quality.
    The evaluation uses a single assessor from the client project (§3). No inter-rater reliability, no second assessor, and no calibrated rubric are reported.

pith-pipeline@v1.3.0-alltime-deepseek · 6261 in / 8957 out tokens · 105650 ms · 2026-08-01T17:01:15.575412+00:00 · methodology

0 comments
Cite this review

Pith. "Pith review of CODENS: Transforming Code Changes into Living, Accessible, and Queryable Documentation." pith.science (2026). https://pith.science/paper/ONC27CM5

@misc{pith2026260718356,
  author       = {Pith},
  title        = {Pith review of: CODENS: Transforming Code Changes into Living, Accessible, and Queryable Documentation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ONC27CM5}},
  note         = {Machine review of arXiv:2607.18356}
}
Share X Bluesky LinkedIn Reddit HN
read the original abstract

Maintaining up-to-date code documentation is difficult in fast-moving repositories because design knowledge is scattered across source files and pull requests. We present CODENS , a system that turns pull requests into living, accessible, and queryable documentation for production codebases. CODENS incrementally builds a typed software knowledge graph from pull requests, enriches components through schema-driven semantic extraction, derives typed relations between them, and exposes the resulting knowledge through three retrieval modes, including agent-guided graph traversal for repository-level question answering. The system also preserves semantic change history across pull requests and integrates both answer-quality and operational evaluation metrics. We evaluate CODENS on a client Ruby on Rails project in production. Results show that CODENS produces highly relevant and well-grounded answers, while qualitative feedback highlights a remaining challenge in concise, documentation-oriented synthesis.

Figures

Figures reproduced from arXiv: 2607.18356 by Abdelhak Kelious, Chyrine Tahri, Eliot Bardet.

Figure 1
Figure 1. Figure 1: CODENS pipeline overview. The system processes pull requests incrementally to build a semantic knowledge graph. it continuously updates a structured project memory with seman￾tic attributes such as purpose, behavioral flow, business logic, and typed inter-component relations. CODENS also combines graph-based and agentic retrieval for documentation-oriented question answering. Microsoft GraphRAG showed that… view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

102 extracted references · 1 canonical work pages

  1. [1]

    Abril and Robert Plant

    Patricia S. Abril and Robert Plant. The patent holder's dilemma: Buy, sell, or troll?. Communications of the ACM. doi:10.1145/1188913.1188915

    arXiv

  2. [2]

    Deciding equivalances among conjunctive aggregate queries

    Sarah Cohen and Werner Nutt and Yehoshua Sagic. Deciding equivalances among conjunctive aggregate queries. doi:10.1145/1219092.1219093

    arXiv

  3. [3]

    Special issue: Digital Libraries. 1996

    1996

  4. [4]

    Understanding Policy-Based Networking

    David Kosiur. Understanding Policy-Based Networking

  5. [7]

    doi:10.1007/3-540-09237-4

    The title of book two. doi:10.1007/3-540-09237-4

    doi:10.1007/3-540-09237-4

  6. [8]

    Asad Z. Spector. Achieving application requirements. Distributed Systems. doi:10.1145/90417.90738

    arXiv

  7. [9]

    Douglass and David Harel and Mark B

    Bruce P. Douglass and David Harel and Mark B. Trakhtenbrot. Statecarts in use: structured analysis and object-orientation. Lectures on Embedded Systems. doi:10.1007/3-540-65193-4_29

    doi:10.1007/3-540-65193-4_29

  8. [10]

    Donald E. Knuth. The Art of Computer Programming, Vol. 1: Fundamental Algorithms (3rd. ed.)

  9. [11]

    Donald E. Knuth. The Art of Computer Programming

  10. [12]

    Structured Variational Inference Procedures and their Realizations (as incol)

    Dan Geiger and Christopher Meek. Structured Variational Inference Procedures and their Realizations (as incol). Proceedings of Tenth International Workshop on Artificial Intelligence and Statistics, The Barbados

  11. [13]

    Stan W. Smith. An experiment in bibliographic mark-up: Parsing metadata for XML export. Proceedings of the 3rd. annual workshop on Librarians and Computers

  12. [14]

    Catch me, if you can: Evading network signatures with web-based polymorphic worms

    Matthew Van Gundy and Davide Balzarotti and Giovanni Vigna. Catch me, if you can: Evading network signatures with web-based polymorphic worms. Proceedings of the first USENIX workshop on Offensive Technologies

  13. [15]

    Predicate Path expressions

    Sten Andler. Predicate Path expressions. Proceedings of the 6th. ACM SIGACT-SIGPLAN symposium on Principles of Programming Languages. doi:10.1145/567752.567774

    arXiv

  14. [16]

    LOGICS of Programs: AXIOMATICS and DESCRIPTIVE POWER

    David Harel. LOGICS of Programs: AXIOMATICS and DESCRIPTIVE POWER

  15. [17]

    Anisi , title =

    David A. Anisi , title =

  16. [18]

    Clarkson

    Kenneth L. Clarkson. Algorithms for Closest-Point Problems (Computational Geometry)

  17. [19]

    Introduction to Bayesian Statistics

    Harry Thornburg. Introduction to Bayesian Statistics. 2001

    2001

  18. [20]

    CLIFFORD: a Maple 11 Package for Clifford Algebra Computations, version 11

    Rafal Ablamowicz and Bertfried Fauser. CLIFFORD: a Maple 11 Package for Clifford Algebra Computations, version 11. 2007

    2007

  19. [21]

    Stats and Analysis

    Poker-Edge.Com. Stats and Analysis. 2006

    2006

  20. [22]

    A more perfect union

    Barack Obama. A more perfect union

  21. [23]

    The fountain of youth

    Joseph Scientist. The fountain of youth

  22. [24]

    Solder man

    Dave Novak. Solder man. ACM SIGGRAPH 2003 Video Review on Animation theater Program: Part I - Vol. 145 (July 27--27, 2003). doi:10.945/woot07-S422

    2003

  23. [25]

    Interview with Bill Kinder: January 13, 2005

    Newton Lee. Interview with Bill Kinder: January 13, 2005. Comput. Entertain. doi:10.1145/1057270.1057278

    arXiv 2005

  24. [26]

    The Enabling of Digital Libraries

    Bernard Rous. The Enabling of Digital Libraries. Digital Libraries

  25. [28]

    (new) Finding minimum congestion spanning trees , journal =

    Werneck, Renato and Setubal, Jo\. (new) Finding minimum congestion spanning trees , journal =. doi:10.1145/351827.384253 , acmid = 384253, publisher =

    arXiv

  26. [30]

    and Mei, Alessandro , title =

    Conti, Mauro and Di Pietro, Roberto and Mancini, Luigi V. and Mei, Alessandro , title =. Inf. Fusion , volume =. 2009 , issn =. doi:10.1016/j.inffus.2009.01.002 , acmid =

    doi:10.1016/j.inffus.2009.01.002 2009

  27. [31]

    and Hutchful, David K

    Li, Cheng-Lun and Buyuktur, Ayse G. and Hutchful, David K. and Sant, Natasha B. and Nainwal, Satyendra K. , title =. CHI '08 extended abstracts on Human factors in computing systems , year =. doi:10.1145/1358628.1358946 , acmid =

    arXiv

  28. [32]

    , title =

    Hollis, Billy S. , title =. 1999 , isbn =

    1999

  29. [33]

    Goossens, Michel and Rahtz, S. P. and Moore, Ross and Sutor, Robert S. , title =. 1999 , isbn =

    1999

  30. [34]

    and Rosenberg, Arnold L

    Buss, Jonathan F. and Rosenberg, Arnold L. and Knott, Judson D. , title =. 1987 , source =

    1987

  31. [35]

    CHI '08: CHI '08 extended abstracts on Human factors in computing systems , year =

    , note =. CHI '08: CHI '08 extended abstracts on Human factors in computing systems , year =

  32. [36]

    Algorithms for Closest-Point Problems (Computational Geometry) , year =

    Clarkson, Kenneth Lee , advisor =. Algorithms for Closest-Point Problems (Computational Geometry) , year =

  33. [37]

    SIGCOMM Comput. Commun. Rev. , year =

  34. [38]

    2004 , isbn =

    IEEE TCSC Executive Committee , booktitle =. 2004 , isbn =. doi:http://dx.doi.org/10.1109/ICWS.2004.64 , acmid =

    doi:10.1109/icws.2004.64 2004

  35. [39]

    Distributed systems (2nd Ed.) , year =

  36. [40]

    , title =

    Petrie, Charles J. , title =. 1986 , source =

    1986

  37. [41]

    Donald E. Knuth. Seminumerical Algorithms. 1981

    1981

  38. [42]

    E-commerce and cultural values , year =

    Kong, Wei-Chang , Title =. E-commerce and cultural values , year =

  39. [43]

    E-commerce and cultural values , year =

    Kong, Wei-Chang , type =. E-commerce and cultural values , year =

  40. [44]

    Chapter 9 , booktitle =

    Kong, Wei-Chang , editor =. Chapter 9 , booktitle =

  41. [45]

    E-commerce and cultural values , editor =

    Kong, Wei-Chang , title =. E-commerce and cultural values , editor =. 2003 , isbn =

    2003

  42. [46]

    E-commerce and cultural values - (InBook-num-in-chap) , chapter =

    Kong, Wei-Chang , editor =. E-commerce and cultural values - (InBook-num-in-chap) , chapter =. 2004 , address =

    2004

  43. [47]

    E-commerce and cultural values (Inbook-text-in-chap) , chapter =

    Kong, Wei-Chang , editor =. E-commerce and cultural values (Inbook-text-in-chap) , chapter =. 2005 , address =

    2005

  44. [48]

    E-commerce and cultural values (Inbook-num chap) , chapter =

    Kong, Wei-Chang , editor =. E-commerce and cultural values (Inbook-num chap) , chapter =. 2006 , address =

    2006

  45. [49]

    Microelectron

    Mehdi Saeedi and Morteza Saheb Zamani and Mehdi Sedighi , title =. Microelectron. J. , volume =. 2010 , pages =

    2010

  46. [50]

    Mehdi Saeedi and Morteza Saheb Zamani and Mehdi Sedighi and Zahra Sasanian , title =. J. Emerg. Technol. Comput. Syst. , volume =

  47. [51]

    Kirschmer, Markus and Voight, John , title =. SIAM J. Comput. , issue_date =. 2010 , issn =. doi:https://doi.org/10.1137/080734467 , acmid =

    doi:10.1137/080734467 2010

  48. [52]

    Hoare, C. A. R. , title =. Structured programming (incoll) , editor =. 1972 , isbn =

    1972

  49. [53]

    History of programming languages I (incoll) , editor =

    Lee, Jan , title =. History of programming languages I (incoll) , editor =. 1981 , isbn =. doi:http://doi.acm.org/10.1145/800025.1198348 , acmid =

    arXiv 1981

  50. [54]

    , title =

    Dijkstra, E. , title =. Classics in software engineering (incoll) , year =

  51. [55]

    , title =

    Wenzel, Elizabeth M. , title =. Multimedia interface design (incoll) , year =. doi:10.1145/146022.146089 , acmid =

    arXiv

  52. [56]

    , title =

    Mumford, E. , title =. Critical issues in information systems research (incoll) , year =

  53. [57]

    and Golden, Donald G

    McCracken, Daniel D. and Golden, Donald G. , title =. 1990 , isbn =

    1990

  54. [58]

    The analysis of linear partial differential operators

    H. The analysis of linear partial differential operators. 1985 , PAGES =

    1985

  55. [59]

    IEEE", address =

    A. Adya and P. Bahl and J. Padhye and A.Wolman and L. Zhou , title =. Proceedings of the IEEE 1st International Conference on Broadnets Networks (BroadNets'04) , publisher = "IEEE", address = "Los Alamitos, CA", year =

  56. [60]

    I. F. Akyildiz and W. Su and Y. Sankarasubramaniam and E. Cayirci , title =. Comm. ACM , volume = 38, number = "4", year =

  57. [61]

    I. F. Akyildiz and T. Melodia and K. R. Chowdhury , title =. Computer Netw. , volume = 51, number = "4", year =

  58. [62]

    ACM", address =

    P. Bahl and R. Chancre and J. Dungeon , title =. Proceeding of the 10th International Conference on Mobile Computing and Networking (MobiCom'04) , publisher = "ACM", address = "New York, NY", year =

  59. [63]

    8 (Special Issue on Sensor Networks)

    D. Culler and D. Estrin and M. Srivastava , title =. IEEE Comput. , volume = 37, number = "8 (Special Issue on Sensor Networks)", publisher = "IEEE", address = "Los Alamitos, CA", year =

  60. [64]

    Natarajan and M

    A. Natarajan and M. Motani and B. de Silva and K. Yap and K. C. Chua , title =. Network Architectures , editor =. 960935712

  61. [65]

    Tzamaloukas and J

    A. Tzamaloukas and J. J. Garcia-Luna-Aceves , title =

  62. [66]

    Zhou and J

    G. Zhou and J. Lu and C.-Y. Wan and M. D. Yarvis and J. A. Stankovic , title =

  63. [67]

    Mapping Powerlists onto Hypercubes

    Jacob Kornerup. Mapping Powerlists onto Hypercubes. 1994

    1994

  64. [68]

    Automatic Parallelization for Distributed-Memory Multiprocessing Systems

    Michael Gerndt. Automatic Parallelization for Distributed-Memory Multiprocessing Systems

  65. [69]

    J. E. Archer, Jr. and R. Conway and F. B. Schneider. User recovery and reversal in interactive systems. ACM Trans. Program. Lang. Syst

  66. [70]

    D. D. Dunlop and V. R. Basili. Generalizing specifications for uniformly implemented loops. ACM Trans. Program. Lang. Syst

  67. [71]

    Heering and P

    J. Heering and P. Klint. Towards monolingual programming environments. ACM Trans. Program. Lang. Syst

  68. [72]

    Donald E. Knuth. The book

  69. [73]

    Korach and D

    E. Korach and D. Rotem and N. Santoro. Distributed algorithms for finding centers and medians in networks. ACM Trans. Program. Lang. Syst

  70. [74]

    : A Document Preparation System

    Leslie Lamport. : A Document Preparation System

  71. [75]

    F. Nielson. Program transformations in a denotational setting. ACM Trans. Program. Lang. Syst

  72. [76]

    Brian K. Reid. A high-level approach to computer document formatting. Proceedings of the 7th Annual Symposium on Principles of Programming Languages

  73. [77]

    and Abdelzaher, Tarek F

    Zhou, Gang and Wu, Yafeng and Yan, Ting and He, Tian and Huang, Chengdu and Stankovic, John A. and Abdelzaher, Tarek F. , title =. ACM Trans. Embed. Comput. Syst. , issue_date =. doi:10.1145/1721695.1721705 , acmid = 1721705, publisher =

    arXiv

  74. [78]

    Institutional members of the Users Group

  75. [79]

    Boris Veytsman , title =

  76. [80]

    Robin Schneider , title =

  77. [81]

    and Peterson, Larry L

    Bowman, Mic and Debray, Saumya K. and Peterson, Larry L. , title =. ACM Trans. Program. Lang. Syst. , volume =. 1993 , doi =

    1993

  78. [82]

    TUGboat , volume =

    Braams, Johannes , title =. TUGboat , volume =

  79. [83]

    Post Congress Tristesse

    Malcolm Clark. Post Congress Tristesse. TeX90 Conference Proceedings

  80. [84]

    ACM Trans

    Herlihy, Maurice , title =. ACM Trans. Program. Lang. Syst. , volume =. 1993 , doi =

    1993

Showing first 80 references.