Login
The f-apps
Login

The source tree contains a number of test/demo apps which are named f-something (the "f" is for "fossil", of course). Collectively, these apps demonstrate the functionality which is currently ported in from fossil(1) (or is new) and working. If a fossil(1) feature is not represented here, it's probably not yet ported. Note that libfossil is still very much in development, so any or all of these may be... fussy... at times. Also note that these are primarily for testing and demonstrating the library, and may not be suitable as apps for every-day use. That said, the active libfossil developers tend to use these applications as "daily drivers" for their fossil work, where appropriate. e.g. f-ci is used for performing most checkins into this repository.

In alphabetical order:

All of these applications accept -?, --help, or a first non-flag argument of help to show their help text.

There are also several apps named f-test-SOMETHING which exist for testing specific APIs and may also be interesting for would-be users of the library.

These applications provide demonstrations of using the library and give devs a place to test new features. fcli provides a mini-framework which takes care of bootstrapping fossil for these applications, making it pretty simple to create new ones. fcli handles the CLI parsing, global flags, opening of a repo/checkout, and other bootstrapping bits. If you've ever hacked on fossil(1), adding a new f-* app is very similar to adding a new command to fossil, with a few more lines of bootstrap code (because each "command" is in its own app).

Demos

# f-wiki ls
Timestamp           UUID     Name
2013-08-22@22:26:59 86afdaa5 AmalgamationBuild
2013-08-23@18:37:26 e451926f building
2013-08-22@14:54:47 042b30df download
2013-08-24@11:29:34 0f083370 f-tools
...

# f-tag -a tip -t demo-tag="demo for prosperity purposes" --trace-sql 
SQL TRACE #1: SELECT '/home/stephan/cvs/fossil/f2' || '/' || value FROM vvar WHERE name='repository';
SQL TRACE #2: ATTACH DATABASE '/home/stephan/cvs/fossil/f2.fsl' AS repo;
SQL TRACE #3: SELECT julianday('now');
SQL TRACE #4: SELECT objid FROM event WHERE type='ci' ORDER BY event.mtime DESC LIMIT 1;
SQL TRACE #5: SELECT uuid FROM blob WHERE rid=1226;
SQL TRACE #6: SELECT strftime('%Y-%m-%dT%H:%M:%f',2456517.117559);
SQL TRACE #7: BEGIN TRANSACTION;
...
SQL TRACE #21: COMMIT;
SQL TRACE #22: DETACH DATABASE repo;


# f-timeline -n 1
g  [daa063582c2c] @ 2013-08-12 16:49:17 by [stephan]
	Edit [2a84ad39]: Add "demo-tag" with value "demo for prosperity purposes".


# f-timeline -n 1 --trace-sql
SQL TRACE #1: SELECT '/home/stephan/cvs/fossil/f2' || '/' || value FROM vvar WHERE name='repository';
SQL TRACE #2: ATTACH DATABASE '/home/stephan/cvs/fossil/f2.fsl' AS repo;
SQL TRACE #3: SELECT substr(uuid,1,12) AS uuid, <...BIG SNIP...> ORDER BY event.mtime DESC LIMIT 1;
g  [daa063582c2c] @ 2013-08-12 16:49:17 by [stephan]
	Edit [2a84ad39]: Add "demo-tag" with value "demo for prosperity purposes".
SQL TRACE #4: DETACH DATABASE repo;


# f-resolve current prev 
b170fb24fe5d94a5b7fa7a00a9911083bf69b27b    1466 current
1ec904fa6a54b45b0a35cd14bab41af138584e45    1461 prev

f-new

f-new can be used to create a new, empty repository, analog to fossil's "new" command. It has one feature fossil(1) does not: it allows the client to specify the commit message for the "initial empty checkin." (Sidebar: the library has a mechanism to disable creation of the initial empty checkin, but fossil/libfossil are largely untested with such repos.)

Example usage:

# f-new -m 'hi there' --config ../fossil.fsl my.fsl --user fred
Copying configuration from: ../fossil.fsl
Created repository: my.fsl
server-code    = a3d6ac6448017382fa1277bbee7e74b96ce385cb
project-code   = e0e38c4a0fb01a2ea277ed8af4e5742d68dbdc60
admin-user     = fred (password=084fdd)

# f-timeline -R my.fsl 
ci [3a661da86ae5] @ 2013-08-28 23:30:39 by [fred] in branch [trunk]
	hi there

# f-acat trunk -R my.fsl 
C hi\sthere
D 2013-08-28T21:30:39.000
R d41d8cd98f00b204e9800998ecf8427e
T *branch * trunk
T *sym-trunk *
U fred
Z 52ef222ec505c27615d80a8552690272

f-tag

f-tag can be used to add and cancel tags on arbitrary repository artifacts. Tags in fossil are actually key/value pairs, but the value is optional. The ability to tag arbitrary artifacts, including other tags, allows (in principal) for applying arbitrary key/value-pair metadata to arbitrary artifacts. This can be used for a great many things above and beyond what fossil currently uses tags for. For example, tags could be used to implement a simple form of comment thread system, by implementing comments as tags, and replies as tags to those comment tags, ad nauseum.

Anyway... this tool is a very basic test/demonstration of this library which allows one to apply tags to artifacts. It allows some things the current fossil UI/CLI cannot make much sense of or do anything useful with (e.g. tags on tags or "cancel" tags with values), but nothing it does is incompatible with how fossil works internally.

FOREWARNINGS: changing branch tags or other "fossil-special" tags (e.g. Wiki pages or tickets) with this tool is as yet untested, is not yet expected to work (it's an ongoing process), and might break things (relationships between data records). Try such things at your own risk. Better yet, don't try them at all until this warning message is gone.

'Private' content which gets tagged will result in the tag being private as well (to avoid that it synchronized without its tagged content).

Examples of setting and cancelling flags:

# Change the color of a single commit:
# f-tag -a current -t bgcolor='#ffff33'

# Propagate the bg color starting at a commit:
# f-tag -a current -t '*bgcolor'='#ffff33'

# Cancel that color:
# f-tag -a current -t=-bgcolor='that was uglier than i thought it would be'

# Change the checkin comment on the current checkout:
# f-tag -a current -t comment='...new comment...'

Values are always optional but Fossil internally treats some tag names specially and may require (or expect) a value. The bgcolor tag is one example. When cancelling a tag, the value is always optional, regardless of whether or not it is a special tag.

When adding a tag, the tag name may optionally be prefixed with a + sign, for symmetry with the cancel (-) and propagate (*) markers. Cancel tags might look like options/flags because they start with a minus, but they are not interpreted as a flag due to a happy accident of design. When using propagating tags, it is wise to enclose the tag in quotes to prevent any unwanted side-effects of shell globbing. Optionally, use -t=*tagname, which would only match a glob in the most unusual of circumstances. Likewise, -t=-tag-to-cancel can be used if having the tag value look like a flag seems disturbing to you.

Note that the +/-/* prefixes were not chosen arbitrarily: they reflect how Fossil internally stores and recognizes the type of a tag: + represents an "add tag", - a "cancel tag" (a.k.a. "anti-tag"), and * a propagating tag.

f-acat

The "artifact cat" tool outputs content from the content. If given a checkin name or the ID of some other fossil-internal structure, it outputs that structure (as opposed to doing complex rendering of the structure-specific type). By default it applies any necessary deltas to produce the desired version of a blob, but it can also be told (using the --raw flag) that it should do no undeltification. It always uncompresses the data (assuming it was compressed, which it normally, but not always, is).

# The current checkout version's manifest:
# f-acat current
B 200d1cd898f4e05591e78f82b2c0f2bc4c8db998
C Added\s--raw,\s--output,\sand\s--artifact\sflags\sto\sf-acat.
D 2013-09-08T12:25:31.660
F Makefile.in 41c00b5876e09e6bef5eaf4a68e885926eeca159
...

# The previous version's manifest, in its raw form (in this case
# a fossil-format delta):
# f-acat prev --raw 
II
i@0,3P:Moved\sfsl_memory_allocator<SNIP>.
D 2013-09-08T12:01:11.6142g@24,9d@5Z,1u:04923c4435ee082205979a2bc0a9b518e1ff3d80
R a5f73027c19c1b33364845d4808bca6d
U stephan
Z 6721ef25586956f13ef59a4dcb77e339

f-query

f-query can be used to run simple SQL commands against a repo/db, kind of like a castrated form of fossil(1)'s sqlite command. The one advantage to using this vs. using the sqlite shell (or fossil(1)) is that this one provides a few fossil-specific SQL functions which those do not. For example, FSL_DIRPART() can be used to figure out the list of directories in a repository:

Example usage:

# f-query -h -e "select distinct(fsl_dirpart(name,1)) dname
    from filename where dname IS NOT NULL order by dname"
.fossil-settings/
.settings/
autosetup/
autosetup/lib/
doc/
...

### Simulate f-acat:
# f-query -h -e "select fsl_content('rid:1')"
C initial\sempty\scheck-in
D 2013-07-23T13:56:36.037
P
R d41d8cd98f00b204e9800998ecf8427e
T *branch * trunk
...

An -e value may either a single SQL statement or the name of a file containing a single SQL statement. Such statements, if they return rows, will have their results dumped to stdout. Contrariwise, the -E flag may be a file or multiple semicolon-separated statements, all of which are executed but none of which will present any results to the user. Both -e and -E may be specified multiple times.

f-wiki

f-wiki can list wiki pages and export/import wiki page content.

Example usage:

# f-wiki ls
Time (local time)    UUID          Size   Name
2013-08-23 00:26:59  86afdaa51885  2470   AmalgamationBuild
2013-08-23 20:37:26  e451926f4c03  1626   building
2013-09-24 19:18:21  a21e5daee961  1827   DbFunctions
...

# f-wiki ls --names
AmalgamationBuild
building
DbFunctions
...

# f-wiki export AmalgamationBuild | head
&lt;h1>The Amalgamation Build&lt;/h1>

See also: [building]
...