pldi-submission

PLDI Submission

Run this audit before the upload, not after. The numbers below were read from the

PLDI 2026 Call for Papers and track pages on 2026-07-08 (via search renderings of

the official URLs — see resources/official-source-map.md); they describe one

cycle and must be rechecked against the live CFP for the cycle you are entering.

The PLDI 2027 call was not yet posted at that date.

The 2026-cycle mechanics

• One deadline a year. PLDI 2026 papers were due Thursday, November 13, 2025.

There is no second round; a missed November costs twelve months.

• HotCRP, not OpenReview. Submission ran at pldi2026.hotcrp.com — profile,

conflicts, and topic areas live there, and each edition gets a fresh instance.

• **Page budget: at most 20 pages of text, excluding bibliography**, in the

single-column ACM acmsmall layout (10 pt font, 12 pt line spacing, text block

5.478 in x 7.884 in), the format PACMPL publishes in. Build with

sample-acmsmall-conf.tex from the acmart package.

• Summary rejection is the stated penalty for a different format — smaller

fonts, a larger text block, squeezed spacing. The chairs do not negotiate

margins.

• Double-blind, and PLDI has used it for years: reviewers must not learn who

you are from the PDF or anything it links to.

Where PL papers leak identity

Compiler and tool papers leak differently from ML papers. The usual culprits: a

tool name already public on GitHub under your lab, benchmark harness URLs,

\author metadata inside the PDF, grant numbers in acknowledgements, and commit

histories inside supplementary archives.

pdfinfo paper.pdf | grep -Ei 'author|creator'
pdftotext paper.pdf - | grep -niE 'github\.com/[^ ]+|gitlab|acknowledg|grant no' | head
unzip -l supplement.zip | grep -Ei '\.git/|/home/|/Users/' | head

Cite your own prior work in the third person, and if the tool is already public,

discuss with your co-authors whether to rename it in the submission or accept the

risk explicitly — do not discover the collision during reviewing.

Summary-rejection and triage table

| Trigger | Consequence at PLDI | Recoverable? |

|---|---|---|

| Modified format (font, text block, spacing) | Summary rejection, stated in the CFP | No, after the deadline |

| Over the text-page cap | Summary rejection risk | No |

| Author identity visible in PDF or links | Desk-level double-blind violation | No |

| Same work under review elsewhere | Dual-submission violation under SIGPLAN/ACM policy | No |

| Evaluation missing baselines or variance | Survives triage, dies in review | Only before submission |

| Artifact unplanned | No submission penalty, but AE starts right after acceptance | Yes, but plan now |

Final week, in order

1. Freeze the claim list; every abstract sentence must have a section that proves it.
2. Rebuild the PDF from a clean tree with the unmodified acmsmall class and

recount text pages against the current cap.

1. Run the leak scan above on the PDF and every supplementary archive.
2. Verify HotCRP conflicts and topics; a stale conflict list misroutes reviewers.
3. Check the CFP's dual-submission and republication wording against anything you

have on arXiv or under review elsewhere.

1. Submit a day early. HotCRP deadlines are hard cutoffs.

What this cycle did not settle (待核实)

Whether an abstract-registration step precedes the paper deadline, exact

supplementary-material rules, and any AI-assistance disclosure requirement were

not confirmed in accessible renderings of the 2026 pages — read the live CFP

rather than assuming either way.

Output format

[PLDI readiness] Ready / Needs fixes / Not ready
[Format] acmsmall unmodified? text pages vs cap?
[Anonymity] PDF metadata / links / tool name / supplement
[Policy] dual submission / arXiv posture / AI disclosure (待核实)
[Fix order] <ordered list before the November deadline>