trivial: remove version-format.md
It's not being kept up to date, and superceded by
https://fwupd.org/lvfs/docs/metainfo/version
Mario Limonciello authored 4 years ago
Richard Hughes committed 4 years ago
| 0 | Version Formats | |
| 1 | =============== | |
| 2 | ||
| 3 | In some circumstances fwupd has to convert from a unsigned integer version | |
| 4 | number into something that has either been used in documentation or has been | |
| 5 | defined in some specification. | |
| 6 | A good example here is the UEFI ESRT table, which specifies a `uint32_t` for | |
| 7 | the version but does not specify how this should be formatted for the user. | |
| 8 | ||
| 9 | As is typical in underspecified specifications, vendors have converted the | |
| 10 | integer in different ways. For instance, Dell uses version strings like 1.2.3 | |
| 11 | and Microsoft use versions like 1.2.3.4. | |
| 12 | ||
| 13 | The fwudp daemon can match specific devices and apply the correct version style | |
| 14 | using quirk files. The version format can also be specified in the firmware | |
| 15 | `metainfo.xml` file so that the new version is correctly shown, and so that it | |
| 16 | matches on the LVFS website. | |
| 17 | ||
| 18 | The current version formats supported by fwupd and the LVFS are: | |
| 19 | ||
| 20 | * `plain`: Use plain integer version numbers with no dots, e.g. `AABBCCDD` | |
| 21 | * `quad`: Use Dell-style `AA.BB.CC.DD` version numbers | |
| 22 | * `triplet`: Use Microsoft-style `AA.BB.CCDD` version numbers | |
| 23 | * `pair`: Use two `AABB.CCDD` version numbers | |
| 24 | * `bcd`: Use binary coded decimal notation | |
| 25 | * `intel-me`: Use Intel ME-style notation (`aaa+11.bbbbb.CC.DDDD`) | |
| 26 | * `intel-me2`: Use alternate Intel ME-style-style `A.B.CC.DDDD` notation | |
| 27 | ||
| 28 | These can be specified in quirk files like this: | |
| 29 | ||
| 30 | # Vendor Modelname | |
| 31 | [Guid=5b92717b-2cad-4a96-a13b-9d65781df8bf] | |
| 32 | VersionFormat = intel-me2 | |
| 33 | ||
| 34 | ...or in metainfo.xml files like this: | |
| 35 | ||
| 36 | <custom> | |
| 37 | <value key="LVFS::VersionFormat">intel-me2</value> | |
| 38 | </custom> | |
| 39 | ||
| 40 | Runtime requirements | |
| 41 | -------------------- | |
| 42 | ||
| 43 | Versions of fwupd `< 1.2.0` can only support firmware updates with key values | |
| 44 | `LVFS::VersionFormat` of `quad` and `triplet`. Additionally, on older versions | |
| 45 | no quirk `VersionFormat` device fixups are supported. | |
| 46 | ||
| 47 | If want to use one of the additional version formats you should depend on a | |
| 48 | specific version of fwupd in the firmware file: | |
| 49 | ||
| 50 | <requires> | |
| 51 | <id compare="ge" version="1.2.0">org.freedesktop.fwupd</id> | |
| 52 | </requires> | |
| 53 | ||
| 54 | This is not *strictly* required, as the integer value can be used for update | |
| 55 | calculations if the version is specified in hex (e.g. `0x12345678`) in the | |
| 56 | `<release>` tag, although the user might get a bit confused if the update | |
| 57 | version does not match the update description. |