[fossil-users] A large repo import suceeded!
I've been using fossil for my personal work for a few months now, but never had the chance to try it on a large project. After seeing the new git import feature, I decided to give it a shot and .. whoa! My work git repo is 2.8GB in size. During the first stage of the import, the fossil file went to about 8.4GB, but after the vacuuming, it shrank to the same 2.8GB. Sweet! Some questions - Q1: Rgd git import - my repo has git-svn branch references of the form svnroot/BranchName. These have now been imported into fossil as branches with name BranchName. Is the svnroot/ info preserved somewhere in the repo? Q2: fossil ui is nice. However, I'm missing search - like what git gui gives. What do fossil-ites use for searching the repo? Just straight database search? Cheers, -Kumar ___ fossil-users mailing list fossil-users@lists.fossil-scm.org http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
Re: [fossil-users] A large repo import suceeded!
On Thu, Dec 02, 2010 at 05:06:03PM +0800, Kumar wrote: My work git repo is 2.8GB in size. During the first stage of the import, the fossil file went to about 8.4GB, but after the vacuuming, it shrank to the same 2.8GB. Sweet! Ok, I gotta ask, what is *in* your repo that it's 2.8 GIGABYTES in size? Is it really 2.8 thousand 1M git repos all screaming to get out? :-) -- Zed A. Shaw http://zedshaw.com/ ___ fossil-users mailing list fossil-users@lists.fossil-scm.org http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
Re: [fossil-users] A large repo import suceeded!
:D I knew someone would ask that! Its not all source (as if I had to say that). There are some media files tracked as well (images, audio clips and video clips) that are part of the shipping product .. and about 3 years worth history of those. -Kumar On Thu, Dec 2, 2010 at 5:08 PM, Zed A. Shaw zeds...@zedshaw.com wrote: On Thu, Dec 02, 2010 at 05:06:03PM +0800, Kumar wrote: My work git repo is 2.8GB in size. During the first stage of the import, the fossil file went to about 8.4GB, but after the vacuuming, it shrank to the same 2.8GB. Sweet! Ok, I gotta ask, what is *in* your repo that it's 2.8 GIGABYTES in size? Is it really 2.8 thousand 1M git repos all screaming to get out? :-) -- Zed A. Shaw http://zedshaw.com/ ___ fossil-users mailing list fossil-users@lists.fossil-scm.org http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
Re: [fossil-users] A large repo import suceeded!
On Thu, Dec 2, 2010 at 4:06 AM, Kumar srikuma...@gmail.com wrote: I've been using fossil for my personal work for a few months now, but never had the chance to try it on a large project. After seeing the new git import feature, I decided to give it a shot and .. whoa! My work git repo is 2.8GB in size. During the first stage of the import, the fossil file went to about 8.4GB, but after the vacuuming, it shrank to the same 2.8GB. Sweet! FWIW, I'm thinking I should rename the import command to git-import - in order to allow future expansion with other import schemes. Similarly, export should be renamed git-export. Or, maybe there is a secondary command to specify the target, for example: fossil import git fossil import svn fossil import hg And so forth Thoughts? Some questions - Q1: Rgd git import - my repo has git-svn branch references of the form svnroot/BranchName. These have now been imported into fossil as branches with name BranchName. Is the svnroot/ info preserved somewhere in the repo? I was not real clear what should be done with those tags, so I stripped off all but the last component of the tag. Is that not the right thing to do? Please explain Q2: fossil ui is nice. However, I'm missing search - like what git gui gives. What do fossil-ites use for searching the repo? Just straight database search? SQLite has a great full-text search engine built in. I've long thought that it would be great to add an interface to this in Fossil. We could index diffs for all check-ins, all wiki, all tickets, all Blog entries, etc, and then have a Google-like interface for searching for things. Of course, the full text index would likely double the size of the repository file, but in this era of TB-size disk drives, is that really an issue? It's really more of a matter of finding the time to do the necessary hacking Cheers, -Kumar ___ fossil-users mailing list fossil-users@lists.fossil-scm.org http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users -- D. Richard Hipp d...@sqlite.org ___ fossil-users mailing list fossil-users@lists.fossil-scm.org http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
Re: [fossil-users] A large repo import suceeded!
On Thu, Dec 02, 2010 at 09:15:55AM -0500, Richard Hipp wrote: FWIW, I'm thinking I should rename the import command to git-import - in order to allow future expansion with other import schemes. Similarly, export should be renamed git-export. Or, maybe there is a secondary command to specify the target, for example: Strictly speaking, it is git-fast-import as that's the input stream it is reading. Either that or git-import looks best too me. Joerg ___ fossil-users mailing list fossil-users@lists.fossil-scm.org http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
Re: [fossil-users] A large repo import suceeded!
On Thu, Dec 02, 2010 at 09:15:55AM -0500, Richard Hipp wrote: Q2: fossil ui is nice. However, I'm missing search - like what git gui gives. What do fossil-ites use for searching the repo? Just straight database search? SQLite has a great full-text search engine built in. I've long thought that it would be great to add an interface to this in Fossil. We could index diffs for all check-ins, all wiki, all tickets, all Blog entries, etc, and then have a Google-like interface for searching for things. Of course, the full text index would likely double the size of the repository file, but in this era of TB-size disk drives, is that really an issue? It's really more of a matter of finding the time to do the necessary hacking I'm happy to hear that! I'd leave that as an option, thought, about doubling or not doubling the repository file. Having the option of a search without indexing, is feasible? ___ fossil-users mailing list fossil-users@lists.fossil-scm.org http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
Re: [fossil-users] A large repo import suceeded!
On Thu, 2 Dec 2010 15:33:26 +0100 Joerg == Joerg Sonnenberger wrote: Joerg Strictly speaking, it is git-fast-import as that's the input Joerg stream it is reading. Either that or git-import looks best too Joerg me. +1 Moreover, I do not think there is need for e.g. git import svn etc. considering that git's fast-import is de-facto standard for exchanging changesets among different DVCS-s. Sincerely, Gour -- Gour | Hlapicina, Croatia | GPG key: CDBF17CA signature.asc Description: PGP signature ___ fossil-users mailing list fossil-users@lists.fossil-scm.org http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
Re: [fossil-users] A large repo import suceeded!
Hello, fossil import git fossil import svn fossil import hg And so forth Thoughts? In my opinion, this is the most correct option for several reasons: 1- Do not pollute the global namespace 2- Make it easier for everyone that is not going to use the option (most of the users), and need to learn fast the fossil suboptions 3- The important actions is import, the detail is from where About the full text search: - Size is important. At least for some uses. It is common to copy fossil repositories as files from one computer to another. At the same time, one repository can have multiple copies. One possibility could be: 1- The first time that the search is going to be used, require the index to be created 2- Create it outside of the fossil repository, either in a directory previously open or in the .fossil or something similar 3- Show all the results Regards, RR 2010/12/2 Richard Hipp d...@sqlite.org: On Thu, Dec 2, 2010 at 4:06 AM, Kumar srikuma...@gmail.com wrote: I've been using fossil for my personal work for a few months now, but never had the chance to try it on a large project. After seeing the new git import feature, I decided to give it a shot and .. whoa! My work git repo is 2.8GB in size. During the first stage of the import, the fossil file went to about 8.4GB, but after the vacuuming, it shrank to the same 2.8GB. Sweet! FWIW, I'm thinking I should rename the import command to git-import - in order to allow future expansion with other import schemes. Similarly, export should be renamed git-export. Or, maybe there is a secondary command to specify the target, for example: fossil import git fossil import svn fossil import hg And so forth Thoughts? Some questions - Q1: Rgd git import - my repo has git-svn branch references of the form svnroot/BranchName. These have now been imported into fossil as branches with name BranchName. Is the svnroot/ info preserved somewhere in the repo? I was not real clear what should be done with those tags, so I stripped off all but the last component of the tag. Is that not the right thing to do? Please explain Q2: fossil ui is nice. However, I'm missing search - like what git gui gives. What do fossil-ites use for searching the repo? Just straight database search? SQLite has a great full-text search engine built in. I've long thought that it would be great to add an interface to this in Fossil. We could index diffs for all check-ins, all wiki, all tickets, all Blog entries, etc, and then have a Google-like interface for searching for things. Of course, the full text index would likely double the size of the repository file, but in this era of TB-size disk drives, is that really an issue? It's really more of a matter of finding the time to do the necessary hacking Cheers, -Kumar ___ fossil-users mailing list fossil-users@lists.fossil-scm.org http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users -- D. Richard Hipp d...@sqlite.org ___ fossil-users mailing list fossil-users@lists.fossil-scm.org http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users ___ fossil-users mailing list fossil-users@lists.fossil-scm.org http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
Re: [fossil-users] A large repo import suceeded!
fossil import git fossil import svn fossil import hg And so forth Thoughts? If the format of the input stream can be auto-detected, then it can simply be fossil import can't it? You'll only need to specify format explicitly for the export I think. Q1: Rgd git import - my repo has git-svn branch references of the form svnroot/BranchName. These have now been imported into fossil as branches with name BranchName. Is the svnroot/ info preserved somewhere in the repo? I was not real clear what should be done with those tags, so I stripped off all but the last component of the tag. Is that not the right thing to do? Please explain Oh! In my case, the git branches named BranchName were tracking the branches named svnroot/BranchName and so I didn't have a problem, but in the general case I can imagine a collision if you just use the last term. As long as both branches are pointing to the same commit, there is no collision issue and you can drop one, but if they aren't then it'll help if the full branch name is retained - like svnroot/BranchName itself. So, ok, fossil-git is not an exact inverse of git-fossil. SQLite has a great full-text search engine built in. I've long thought that it would be great to add an interface to this in Fossil. We could index diffs for all check-ins, all wiki, all tickets, all Blog entries, etc, and then have a Google-like interface for searching for things. Of course, the full text index would likely double the size of the repository file, but in this era of TB-size disk drives, is that really an issue? Such a search interface would be awesome! I don't think the size increase is an issue at all. The only case may be backups, but even they can be done using cloning - that would mean that the search index shouldn't sync. Cheers, -Kumar ___ fossil-users mailing list fossil-users@lists.fossil-scm.org http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
Re: [fossil-users] A large repo import suceeded!
On Thu, Dec 2, 2010 at 4:36 PM, Ramon Ribó ram...@compassis.com wrote: About the full text search: - Size is important. At least for some uses. It is common to copy fossil repositories as files from one computer to another. At the same time, one repository can have multiple copies. One possibility could be: ... 2- Create it outside of the fossil repository, either in a directory previously open or in the .fossil or something similar A related option: allow a FTS only a checked-out copy of the tree, where we already have extra files like _FOSSIL_. Build the FTS during 'fossil open', update it during 'update' ops, and nuke it during 'fossil close'? -- - stephan beal http://wanderinghorse.net/home/stephan/ ___ fossil-users mailing list fossil-users@lists.fossil-scm.org http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
Re: [fossil-users] A large repo import suceeded!
On Thu, Dec 2, 2010 at 10:21 AM, Gour g...@atmarama.net wrote: On Thu, 2 Dec 2010 15:33:26 +0100 Joerg == Joerg Sonnenberger wrote: Joerg Strictly speaking, it is git-fast-import as that's the input Joerg stream it is reading. Either that or git-import looks best too Joerg me. +1 Moreover, I do not think there is need for e.g. git import svn etc. considering that git's fast-import is de-facto standard for exchanging changesets among different DVCS-s. There are limitations with fast-import. For example, when Fossil exports to the fast-import format, it has to omit all information about wiki and tickets and blog entries, because there is no provision for that in the format. So I'm thinking that fast-import is not the final word in DVCS interchange formats. Fossil should be prepared for the next interchange format, whatever that format might turn out to be Sincerely, Gour -- Gour | Hlapicina, Croatia | GPG key: CDBF17CA ___ fossil-users mailing list fossil-users@lists.fossil-scm.org http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users -- D. Richard Hipp d...@sqlite.org ___ fossil-users mailing list fossil-users@lists.fossil-scm.org http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
Re: [fossil-users] A large repo import suceeded!
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 On 12/02/2010 09:36 AM, Ramon Ribó wrote: In my opinion, this is the most correct option for several reasons: 1- Do not pollute the global namespace 2- Make it easier for everyone that is not going to use the option (most of the users), and need to learn fast the fossil suboptions 3- The important actions is import, the detail is from where I agree with this for all the reasons stated. If import/export will only do git, then it'd make sense to rename them. If they may do other formats in the future, then I think that a subcommand would be best. One thing that I really like about Fossil is that fossil help fits on a single screen and includes *all* commands (other than the test ones that aren't really needed anyway), whereas git help uses two screens and only includes the most common commands. And even then, git help doesn't include some commands I regularly need. I realize that's mostly cosmetic, and I won't spend much time painting this particular bikeshed, but import/export aren't common operations, and having a bunch of top-level commands just for that purpose would seem to crowd that display even further and introduce command creep. -BEGIN PGP SIGNATURE- Version: GnuPG v1.4.10 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/ iEYEARECAAYFAkz3wlUACgkQIaMjFWMehWJFZgCbB7TbEFzcBW5Y4jDJC9/VA4Cj GxMAn2VmnODlfSDRMqMK8Q/UQtfhPhB0 =jS1j -END PGP SIGNATURE- ___ fossil-users mailing list fossil-users@lists.fossil-scm.org http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
Re: [fossil-users] A large repo import suceeded!
SQLite has a great full-text search engine built in. I've long thought that it would be great to add an interface to this in Fossil. We could index diffs for all check-ins, all wiki, all tickets, all Blog entries, etc, and then have a Google-like interface for searching for things. Of course, the full text index would likely double the size of the repository file, but in this era of TB-size disk drives, is that really an issue? Such a search interface would be awesome! I don't think the size increase is an issue at all. The only case may be backups, but even they can be done using cloning - that would mean that the search index shouldn't sync. Because fossil stores artifacts as zlib compressed deltas, full text searching using the existing fossil SQLite database is not possible. One would either have to create a separate database containing the full set of artifacts or create a program to reconstruct each artifact of the database and do a search on each artifact separately. Does anybody have a better idea? David Shinn ___ fossil-users mailing list fossil-users@lists.fossil-scm.org http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
Re: [fossil-users] A large repo import suceeded!
David Shinn wrote: Because fossil stores artifacts as zlib compressed deltas, full text searching using the existing fossil SQLite database is not possible. One would either have to create a separate database containing the full set of artifacts or create a program to reconstruct each artifact of the database and do a search on each artifact separately. Does anybody have a better idea? Repo search is really two problems I think - a) Search the artifacts - i.e content - with revision specified. This can be equivalent in cost to checking out a version and running a file grep. b) Searching the commit logs. Maybe only the commit logs can be stored uncompressed (maybe indexed as well) in order to enable some searching. That'll get us search by text/author/date/files-changed. Thots? ___ fossil-users mailing list fossil-users@lists.fossil-scm.org http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users