BML examples: HTML vs BML
Every example page with both versions. All numbers in bytes, measured by tools/bml_examples.sh on 2026-10-24 14:57 MDT.
| Page | Links | HTML size | BML size | Size diff | HTML gzip -9 | BML gzip -9 | HTTP transfer | bml:// transfer | Transfer diff | HTTP page total | bml:// page total | Total diff | Resources |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| pages/article | html bml | 1795 | 1279 | 516 (28.7%) | 930 | 825 | 2129 | 1433 | 696 (32.7%) | 2866 | 1871 | 995 (34.7%) | demo.css placeholder.png* |
| pages/forms | html bml | 2218 | 1366 | 852 (38.4%) | 992 | 839 | 2550 | 1407 | 1143 (44.8%) | 2891 | 1449 | 1442 (49.9%) | demo.css |
| pages/i18n | html bml | 1655 | 1292 | 363 (21.9%) | 1008 | 960 | 1986 | 1332 | 654 (32.9%) | 2327 | 1374 | 953 (41.0%) | demo.css |
| pages/iframe | html bml | 935 | 720 | 215 (23.0%) | 493 | 451 | 1267 | 762 | 505 (39.9%) | 4151 | 3347 | 804 (19.4%) | demo.css typography.html* |
| pages/index | html bml | 1839 | 1523 | 316 (17.2%) | 644 | 637 | 2171 | 1564 | 607 (28.0%) | 2512 | 1606 | 906 (36.1%) | demo.css |
| pages/interactive | html bml | 1340 | 999 | 341 (25.4%) | 684 | 625 | 1678 | 1046 | 632 (37.7%) | 2019 | 1088 | 931 (46.1%) | demo.css |
| pages/links | html bml | 1411 | 1097 | 314 (22.3%) | 707 | 666 | 1743 | 1138 | 605 (34.7%) | 2084 | 1180 | 904 (43.4%) | demo.css |
| pages/lists | html bml | 1326 | 1043 | 283 (21.3%) | 705 | 665 | 1658 | 1084 | 574 (34.6%) | 1999 | 1126 | 873 (43.7%) | demo.css |
| pages/media | html bml | 2184 | 1612 | 572 (26.2%) | 986 | 903 | 2516 | 1653 | 863 (34.3%) | 4076 | 2586 | 1490 (36.6%) | demo.css flower.vtt placeholder.png* placeholder.svg* |
| pages/newtags | html bml | 1548 | 1005 | 543 (35.1%) | 778 | 665 | 1882 | 1048 | 834 (44.3%) | 2223 | 1090 | 1133 (51.0%) | demo.css |
| pages/tables | html bml | 1388 | 959 | 429 (30.9%) | 665 | 582 | 1721 | 1001 | 720 (41.8%) | 2062 | 1043 | 1019 (49.4%) | demo.css |
| pages/typography | html bml | 2206 | 1845 | 361 (16.4%) | 1322 | 1234 | 2543 | 1891 | 652 (25.6%) | 2884 | 1933 | 951 (33.0%) | demo.css |
| pages/videos | html bml | 2214 | 1677 | 537 (24.3%) | 621 | 557 | 2547 | 1719 | 828 (32.5%) | 3284 | 2157 | 1127 (34.3%) | demo.css placeholder.png* |
| pages7/cjk | html bml | 567 | 490 | 77 (13.6%) | 475 | 453 | 897 | 641 | 256 (28.5%) | 1293 | 1037 | 256 (19.8%) | placeholder.png* |
| pages7/video1 | html bml | 383 | 285 | 98 (25.6%) | 287 | 251 | 716 | 328 | 388 (54.2%) | 3201073 | 3200685 | 388 (0.0%) | clip1.mp4* |
| pages7/video4 | html bml | 593 | 394 | 199 (33.6%) | 292 | 260 | 926 | 437 | 489 (52.8%) | 20570876 | 20570387 | 489 (0.0%) | clip1.mp4* clip2.mp4* clip3.mp4* clip4.mp4* |
| Total | 23602 | 17586 | 6016 (25.5%) | 11589 | 10573 | 28930 | 18484 | 10446 (36.1%) | 23808620 | 23793959 | 14661 (0.1%) |
How each number was measured
- Size: wc -c of the file. Size diff: HTML minus BML, and as a percent of HTML.
- gzip -9: gzip -9 -c file | wc -c (content only, no transfer).
- HTTP transfer: real local fetch from bmlserver (HTTP side) with curl; request + response headers + body (curl size_request + size_header + size_download). No gzip encoding was negotiated.
- bml:// transfer: BlitzBrowse7 fetched bml://127.0.0.1 from the same bmlserver through tools/bytecount_proxy, which counts the real TCP payload bytes both ways (request frame + response frame). Loopback; TCP/IP headers are not counted on either side.
- Page total: page transfer plus each embedded local resource (stylesheet, img/video/track src, poster), each fetched the same way over its protocol. External https resources are not counted. Resources marked * could not be fetched over bml:// (bmlserver answered with a short error frame instead of the file), so for those the HTTP bytes are counted on both sides; the bml:// page-total column therefore mixes bml:// page bytes with HTTP resource bytes for starred resources.