Mercurial > hg4j
changeset 168:dd525ca65de8
Stupid defect that cost me couple of hours. What a coincidence for merge revision of interest (157) to be right on the BIS buffer boundary.
| author | Artem Tikhomirov <tikhomirov.artem@gmail.com> |
|---|---|
| date | Tue, 22 Mar 2011 21:18:40 +0100 |
| parents | 86f9581f4301 |
| children | 8c8e3f372fa1 |
| files | src/org/tmatesoft/hg/internal/RevlogDump.java |
| diffstat | 1 files changed, 4 insertions(+), 1 deletions(-) [+] |
line wrap: on
line diff
--- a/src/org/tmatesoft/hg/internal/RevlogDump.java Tue Mar 22 03:52:34 2011 +0100 +++ b/src/org/tmatesoft/hg/internal/RevlogDump.java Tue Mar 22 21:18:40 2011 +0100 @@ -73,7 +73,10 @@ int parent2Revision = di.readInt(); byte[] buf = new byte[32]; di.readFully(buf, 12, 20); - dis.skip(12); + dis.skipBytes(12); + // CAN'T USE skip() here without extra precautions. E.g. I ran into situation when + // buffer was 8192 and BufferedInputStream was at position 8182 before attempt to skip(12). + // BIS silently skips available bytes and leaves me two extra bytes that ruin the rest of the code. System.out.printf("%4d:%14d %6X %10d %10d %10d %10d %8d %8d %040x\n", entryCount, offset, flags, compressedLen, actualLen, baseRevision, linkRevision, parent1Revision, parent2Revision, new BigInteger(buf)); if (inlineData) { String resultString;