Bug: bea/12c
ID : | 12c986be-d19a-4b8b-b1b5-68248ff4d331 |
Short name : | bea/12c |
Status : | unconfirmed |
Severity : | wishlist |
Assigned : | |
Reporter : | Ronny Pfannschmidt <Ronny.Pfannschmidt@gmx.de> |
Creator : | W. Trevor King <wking@drexel.edu> |
Created : | Tue, 21 Jul 2009 18:32:12 +0000 |
Summary : | Bug aggregation. Multi-repo meta-BE? |
Comment: |
--------- Comment --------- ID: 13012b22-2d02-444c-87c0-8cf0f17137ae Short name: bea/12c/130 From: W. Trevor King <wking@drexel.edu> Date: Sat, 11 Jul 2009 08:50:30 -0400 On Sat, Jul 11, 2009 at 01:54:54PM +0200, Ronny Pfannschmidt wrote: > 1. is there any way to aggregate over multiple public branches in order > to get the complete bug state Keeping the bug data with the source helps synchronize bug state and source code. Bug state in branch A may not apply to branch B. Some people like to weaken this source-bug linkage by keeping their bugs in a branch all by themselves (ditz [http://ditz.rubyforge.org/] currently supports this workflow). It sounds like you want to move from "bugs with code" to "bugs and code in separate branches". We don't have an easy way to do that in BE at the moment, since version-control systems like Git have a single working branch at a time (I think :p). What VCS are you using as a backend? > 2. is there any model for storing bigger files at a central place (for > some of my bugs i have multi-megabyte tarballs attached) be comment ID "See the tarball at http://yourpage/something.tar.gz" Then to grab the tarball, you'd use: wget `be show COMMENT-ID | sed -n 's/ *See the tarball at //p'` to grab it. -- This email may be signed or encrypted with GPG (http://www.gnupg.org). The GPG signature (if present) will be attached as 'signature.asc'. For more information, see http://en.wikipedia.org/wiki/Pretty_Good_Privacy My public key is at http://www.physics.drexel.edu/~wking/pubkey.txt |
Comment: |
--------- Comment --------- ID: dc32aa62-cf56-4171-84a1-8f7d02b23b6d Short name: bea/12c/dc3 From: Ronny Pfannschmidt <Ronny.Pfannschmidt@gmx.de> Date: Sat, 11 Jul 2009 15:13:05 +0200 On Sat, 2009-07-11 at 08:50 -0400, W. Trevor King wrote: > On Sat, Jul 11, 2009 at 01:54:54PM +0200, Ronny Pfannschmidt wrote: > > 1. is there any way to aggregate over multiple public branches in order > > to get the complete bug state > > Keeping the bug data with the source helps synchronize bug state and > source code. Bug state in branch A may not apply to branch B. Some > people like to weaken this source-bug linkage by keeping their bugs in > a branch all by themselves (ditz [http://ditz.rubyforge.org/] > currently supports this workflow). It sounds like you want to move > from "bugs with code" to "bugs and code in separate branches". We > don't have an easy way to do that in BE at the moment, since > version-control systems like Git have a single working branch at a > time (I think :p). What VCS are you using as a backend? the basic idea is to take a look at all public branches (for exaple all on lp/bitbucket/github) in order to tell the user of a webinterface that bug foo is fixed in branch xyz, and if its merged to the main branch > > > 2. is there any model for storing bigger files at a central place (for > > some of my bugs i have multi-megabyte tarballs attached) > > be comment ID "See the tarball at http://yourpage/something.tar.gz" > Then to grab the tarball, you'd use: > wget `be show COMMENT-ID | sed -n 's/ *See the tarball at //p'` > to grab it. so the basic idea is to do it completely self-managed and have have heterogenous sources of extended data? |
Comment: |
--------- Comment --------- ID: 1f9f60de-ba37-42bc-a1c0-dc062ef255e1 Short name: bea/12c/1f9 From: Ben Finney <bignose+hates-spam@benfinney.id.au> Date: Sat, 11 Jul 2009 23:31:34 +1000 Ronny Pfannschmidt <Ronny.Pfannschmidt@gmx.de> writes: > 1. is there any way to aggregate over multiple public branches in > order to get the complete bug state The bug state is as complete as the source code state. It's exactly as aggregated as the rest of the source code; the “complete bug state” would be the integration branch where you merge all the feature branches and bug-fix branches together. If instead you want bugs to *not* be tightly linked with the rest of the source code state, it seems you don't want a distributed bug tracker like Bugs Everywhere. -- \ “I cannot conceive that anybody will require multiplications at | `\ the rate of 40,000 or even 4,000 per hour …” —F. H. Wales, 1936 | _o__) | Ben Finney _______________________________________________ Be-devel mailing list Be-devel@bugseverywhere.org http://void.printf.net/cgi-bin/mailman/listinfo/be-devel |
Comment: |
--------- Comment --------- ID: fd6162f3-7fc1-41d1-a073-a07465802b72 Short name: bea/12c/fd6 From: Gianluca Montecchi <gian@grys.it> Date: Mon, 13 Jul 2009 11:03:41 +0200 On Sat, Jul 11, 2009 at 11:31:34PM +1000, Ben Finney wrote: > Ronny Pfannschmidt <Ronny.Pfannschmidt@gmx.de> writes: > > > 1. is there any way to aggregate over multiple public branches in > > order to get the complete bug state > > The bug state is as complete as the source code state. It's exactly as > aggregated as the rest of the source code; the ???complete bug state??? > would be the integration branch where you merge all the feature branches > and bug-fix branches together. > > If instead you want bugs to *not* be tightly linked with the rest of the > source code state, it seems you don't want a distributed bug tracker > like Bugs Everywhere. "the complete bug state" probably means that he want to know (and in some way to publish it) that the bug "xyz" is fixed and merged in main while bug "abc" is fixed but only in branch "123" and bug "def" is still open in branch "456" bye Gianluca _______________________________________________ Be-devel mailing list Be-devel@bugseverywhere.org http://void.printf.net/cgi-bin/mailman/listinfo/be-devel |
Comment: |
--------- Comment --------- ID: 0f60a148-7024-44bd-bbed-377cbece9d1b Short name: bea/12c/0f6 From: Ben Finney <bignose+hates-spam@benfinney.id.au> Date: Sat, 11 Jul 2009 23:34:08 +1000 Ronny Pfannschmidt <Ronny.Pfannschmidt@gmx.de> writes: > the basic idea is to take a look at all public branches (for exaple > all on lp/bitbucket/github) in order to tell the user of a > webinterface that bug foo is fixed in branch xyz, and if its merged to > the main branch I don't understand. The state of the bug in the main branch is right there in the main branch; if it's not fixed there, it's not fixed there. If it's merged in from a different branch, the bug state follows all the other changes when they come in. Can you give an example of what would be done differently? -- \ “The basic fact about human existence is not that it is a | `\ tragedy, but that it is a bore.” —Henry L. Mencken | _o__) | Ben Finney _______________________________________________ Be-devel mailing list Be-devel@bugseverywhere.org http://void.printf.net/cgi-bin/mailman/listinfo/be-devel |
Comment: |
--------- Comment --------- ID: d86e497d-667d-4c2b-9249-76026df56633 Short name: bea/12c/d86 From: Ronny Pfannschmidt <Ronny.Pfannschmidt@gmx.de> Date: Sat, 11 Jul 2009 16:00:57 +0200 On Sat, 2009-07-11 at 23:34 +1000, Ben Finney wrote: > Ronny Pfannschmidt <Ronny.Pfannschmidt@gmx.de> writes: > > > the basic idea is to take a look at all public branches (for exaple > > all on lp/bitbucket/github) in order to tell the user of a > > webinterface that bug foo is fixed in branch xyz, and if its merged to > > the main branch > > I don't understand. The state of the bug in the main branch is right > there in the main branch; if it's not fixed there, it's not fixed there. > If it's merged in from a different branch, the bug state follows all the > other changes when they come in. > > Can you give an example of what would be done differently? > i want to see the combination of the bug data of all branches for example i got bug its fixed in the branch "something" its not fixed/merged to "main" now something like a website should tell me, this bug has been fixed in branch xyz and the fix is not yet merged into main _______________________________________________ Be-devel mailing list Be-devel@bugseverywhere.org http://void.printf.net/cgi-bin/mailman/listinfo/be-devel |
Comment: |
--------- Comment --------- ID: e520239c-8d69-4ff6-b1bd-0c2f74366200 Short name: bea/12c/e52 From: Ben Finney <bignose+hates-spam@benfinney.id.au> Date: Sun, 12 Jul 2009 00:57:35 +1000 Ronny Pfannschmidt <Ronny.Pfannschmidt@gmx.de> writes: > i want to see the combination of the bug data of all branches What is your definition of “all branches”? When I'm working with distributed VCS, I create branches wherever I feel like, and the VCS tool doesn't have some central registry of branches to keep up to date. How is a tool to determine the set of “all branches”? The distributed VCS model means that set is indeterminate. -- \ “Pinky, are you pondering what I'm pondering?” “I think so, | `\ Brain, but I find scratching just makes it worse.” —_Pinky and | _o__) The Brain_ | Ben Finney _______________________________________________ Be-devel mailing list Be-devel@bugseverywhere.org http://void.printf.net/cgi-bin/mailman/listinfo/be-devel |
Comment: |
--------- Comment --------- ID: bd98f525-95ec-446a-84e8-34c7d6fa5b40 Short name: bea/12c/bd9 From: W. Trevor King <wking@drexel.edu> Date: Sat, 11 Jul 2009 11:25:07 -0400 On Sat, Jul 11, 2009 at 03:13:05PM +0200, Ronny Pfannschmidt wrote: > On Sat, 2009-07-11 at 08:50 -0400, W. Trevor King wrote: > > On Sat, Jul 11, 2009 at 01:54:54PM +0200, Ronny Pfannschmidt wrote: > > > 1. is there any way to aggregate over multiple public branches in order > > > to get the complete bug state > > > > Keeping the bug data with the source helps synchronize bug state and > > source code. Bug state in branch A may not apply to branch B. Some > > people like to weaken this source-bug linkage by keeping their bugs in > > a branch all by themselves (ditz [http://ditz.rubyforge.org/] > > currently supports this workflow). It sounds like you want to move > > from "bugs with code" to "bugs and code in separate branches". We > > don't have an easy way to do that in BE at the moment, since > > version-control systems like Git have a single working branch at a > > time (I think :p). What VCS are you using as a backend? > > the basic idea is to take a look at all public branches (for exaple all > on lp/bitbucket/github) in order to tell the user of a webinterface that > bug foo is fixed in branch xyz, and if its merged to the main branch Hmm. > > > 2. is there any model for storing bigger files at a central place (for > > > some of my bugs i have multi-megabyte tarballs attached) > > > > be comment ID "See the tarball at http://yourpage/something.tar.gz" > > Then to grab the tarball, you'd use: > > wget `be show COMMENT-ID | sed -n 's/ *See the tarball at //p'` > > to grab it. > so the basic idea is to do it completely self-managed Well, it's going to be managed by somebody ;). So far I'm not convinced enough for the manager to be me, so I'm suggesting it be you :p. > and have have heterogenous sources of extended data? I assume "extended data" here refers to your tarballs. What sort of homogenous source did you have in mind? The comment body is currently just a binary blob for non-text/* types, otherwise it's text in whatever encoding you've configured. On Sun, Jul 12, 2009 at 12:57:35AM +1000, Ben Finney wrote: > Ronny Pfannschmidt <Ronny.Pfannschmidt@gmx.de> writes: > > > i want to see the combination of the bug data of all branches > > How is a tool to determine the set of “all branches”? The distributed > VCS model means that set is indeterminate. He could just make a list of branches he likes. Ronny, are you looking to check bug status across several repos on the fly, or periodically run something (with cron, etc.) to update a static multi-repo summary? The easiest implementation I can think of would be to keep local branches (on whatever computer is hosting your web interface) following your favorite repos. proxectX/ |-- repoA |-- repoB `-- repoC You'd pull upstream changes with a cron job. Listing bugs would be something along the lines of projectX$ for repo in * do pushd $repo be list popd done | sort | uniq Then to show bug status you would have something like projectX$ for repo in * do echo $repo pushd $repo be show ${BUGID} popd done For a web frontend, you'd want to translate that to python/libbe. -- This email may be signed or encrypted with GPG (http://www.gnupg.org). The GPG signature (if present) will be attached as 'signature.asc'. For more information, see http://en.wikipedia.org/wiki/Pretty_Good_Privacy My public key is at http://www.physics.drexel.edu/~wking/pubkey.txt |
Comment: |
--------- Comment --------- ID: 8ffc90d7-0be7-4b00-88e6-9ae1b65f7957 Short name: bea/12c/8ff From: Ronny Pfannschmidt <Ronny.Pfannschmidt@gmx.de> Date: Sun, 12 Jul 2009 23:20:10 +0200 On Sat, 2009-07-11 at 11:25 -0400, W. Trevor King wrote: > On Sat, Jul 11, 2009 at 03:13:05PM +0200, Ronny Pfannschmidt wrote: > > On Sat, 2009-07-11 at 08:50 -0400, W. Trevor King wrote: > > > On Sat, Jul 11, 2009 at 01:54:54PM +0200, Ronny Pfannschmidt wrote: > > > > 1. is there any way to aggregate over multiple public branches in order > > > > to get the complete bug state > > > > > > Keeping the bug data with the source helps synchronize bug state and > > > source code. Bug state in branch A may not apply to branch B. Some > > > people like to weaken this source-bug linkage by keeping their bugs in > > > a branch all by themselves (ditz [http://ditz.rubyforge.org/] > > > currently supports this workflow). It sounds like you want to move > > > from "bugs with code" to "bugs and code in separate branches". We > > > don't have an easy way to do that in BE at the moment, since > > > version-control systems like Git have a single working branch at a > > > time (I think :p). What VCS are you using as a backend? > > > > the basic idea is to take a look at all public branches (for exaple all > > on lp/bitbucket/github) in order to tell the user of a webinterface that > > bug foo is fixed in branch xyz, and if its merged to the main branch > > Hmm. > > > > > 2. is there any model for storing bigger files at a central place (for > > > > some of my bugs i have multi-megabyte tarballs attached) > > > > > > be comment ID "See the tarball at http://yourpage/something.tar.gz" > > > Then to grab the tarball, you'd use: > > > wget `be show COMMENT-ID | sed -n 's/ *See the tarball at //p'` > > > to grab it. > > so the basic idea is to do it completely self-managed > > Well, it's going to be managed by somebody ;). So far I'm not > convinced enough for the manager to be me, so I'm suggesting it be you > :p. > > > and have have heterogenous sources of extended data? > > I assume "extended data" here refers to your tarballs. What sort of > homogenous source did you have in mind? The comment body is currently > just a binary blob for non-text/* types, otherwise it's text in > whatever encoding you've configured. some kind of common upload target for a single project in order to have more reliable sources of stuff thats related to bugs but doesnt fit into the normal repository > > On Sun, Jul 12, 2009 at 12:57:35AM +1000, Ben Finney wrote: > > Ronny Pfannschmidt <Ronny.Pfannschmidt@gmx.de> writes: > > > > > i want to see the combination of the bug data of all branches > > > > How is a tool to determine the set of “all branches”? The distributed > > VCS model means that set is indeterminate. > > He could just make a list of branches he likes. > > Ronny, are you looking to check bug status across several repos on the > fly, or periodically run something (with cron, etc.) to update a > static multi-repo summary? on the fly access > > The easiest implementation I can think of would be to keep local > branches (on whatever computer is hosting your web interface) > following your favorite repos. > proxectX/ > |-- repoA > |-- repoB > `-- repoC > You'd pull upstream changes with a cron job. > Listing bugs would be something along the lines of > projectX$ for repo in * > do > pushd $repo > be list > popd > done | sort | uniq > Then to show bug status you would have something like > projectX$ for repo in * > do > echo $repo > pushd $repo > be show ${BUGID} > popd > done > For a web frontend, you'd want to translate that to python/libbe. > yes, the idea is to get a web fontend for multiple branches and maybe a local gtk fontend for local multi-branch setups for some of my projects i have n local and m remote repos, and merging is not always intended soonish |
Comment: |
--------- Comment --------- ID: 4d192c6c-a4a8-4844-b083-2dd5926bd2d9 Short name: bea/12c/4d1 From: W. Trevor King <wking@drexel.edu> Date: Sun, 12 Jul 2009 19:55:02 -0400 On Sun, Jul 12, 2009 at 11:20:10PM +0200, Ronny Pfannschmidt wrote: > On Sat, 2009-07-11 at 11:25 -0400, W. Trevor King wrote: > > On Sat, Jul 11, 2009 at 03:13:05PM +0200, Ronny Pfannschmidt wrote: > > > On Sat, 2009-07-11 at 08:50 -0400, W. Trevor King wrote: > > > > On Sat, Jul 11, 2009 at 01:54:54PM +0200, Ronny Pfannschmidt wrote: > > > > > 2. is there any model for storing bigger files at a central place (for > > > > > some of my bugs i have multi-megabyte tarballs attached) > > > > > > > > be comment ID "See the tarball at http://yourpage/something.tar.gz" > > > > Then to grab the tarball, you'd use: > > > > wget `be show COMMENT-ID | sed -n 's/ *See the tarball at //p'` > > > > to grab it. > > > > > > so the basic idea is to do it completely self-managed > > > and have have heterogenous sources of extended data? > > > > I assume "extended data" here refers to your tarballs. What sort of > > homogenous source did you have in mind? The comment body is currently > > just a binary blob for non-text/* types, otherwise it's text in > > whatever encoding you've configured. > > some kind of common upload target for a single project in order to have > more reliable sources of stuff thats related to bugs but doesnt fit into > the normal repository Sorry, I'm still having trouble with "doesn't fit into the normal repository". It's going to be large wherever you keep it. You worried about multiple branches all having these big tarballs in them and want a "lightweight" checkout without all the big tarballs/whatever? I still think having some sort of "resources" directory on an http server somewhere that you link to from comments is the best plan. If you use the be show --xml ID | be-xml-to-mbox | catmutt approach, you can even write your comments in text/html and get clickable links ;). A "push big file to remote and commit comment linking to it" script would be pretty simple and keep everything consistent. > > On Sun, Jul 12, 2009 at 12:57:35AM +1000, Ben Finney wrote: > > > Ronny Pfannschmidt <Ronny.Pfannschmidt@gmx.de> writes: > > > > > > > i want to see the combination of the bug data of all branches > > > > > > How is a tool to determine the set of “all branches”? The distributed > > > VCS model means that set is indeterminate. > > > > He could just make a list of branches he likes. > > > > Ronny, are you looking to check bug status across several repos on the > > fly, or periodically run something (with cron, etc.) to update a > > static multi-repo summary? > > on the fly access Then listing bugs in a remote repo will either involve httping tons of tiny values files for each bug (slow?) or running some hypothetical BE-server locally for each repo speaking some BE-specific protocol (complicated?). And how would you handle e.g. headless git repos, where nothing's even checked out? You could always run the cron job every 15 minutes, and rely on whatever VCS you're using having some intelligent protocol/procedure to keep bandwidth down ;). If you need faster / more-efficient updates, you'll probably need to throw out polling altogether and setup all involved repos with a "push to central-repo on commit" hook or some such. -- This email may be signed or encrypted with GPG (http://www.gnupg.org). The GPG signature (if present) will be attached as 'signature.asc'. For more information, see http://en.wikipedia.org/wiki/Pretty_Good_Privacy My public key is at http://www.physics.drexel.edu/~wking/pubkey.txt |
Comment: |
--------- Comment --------- ID: 6dcc910a-ce15-4eeb-b49b-4747719748ed Short name: bea/12c/6dc From: Ronny Pfannschmidt <Ronny.Pfannschmidt@gmx.de> Date: Mon, 13 Jul 2009 09:05:34 +0200 On Sun, 2009-07-12 at 19:55 -0400, W. Trevor King wrote: > On Sun, Jul 12, 2009 at 11:20:10PM +0200, Ronny Pfannschmidt wrote: > > On Sat, 2009-07-11 at 11:25 -0400, W. Trevor King wrote: > > > On Sat, Jul 11, 2009 at 03:13:05PM +0200, Ronny Pfannschmidt wrote: > > > > On Sat, 2009-07-11 at 08:50 -0400, W. Trevor King wrote: > > > > > On Sat, Jul 11, 2009 at 01:54:54PM +0200, Ronny Pfannschmidt wrote: > > > > > > 2. is there any model for storing bigger files at a central place (for > > > > > > some of my bugs i have multi-megabyte tarballs attached) > > > > > > > > > > be comment ID "See the tarball at http://yourpage/something.tar.gz" > > > > > Then to grab the tarball, you'd use: > > > > > wget `be show COMMENT-ID | sed -n 's/ *See the tarball at //p'` > > > > > to grab it. > > > > > > > > so the basic idea is to do it completely self-managed > > > > and have have heterogenous sources of extended data? > > > > > > I assume "extended data" here refers to your tarballs. What sort of > > > homogenous source did you have in mind? The comment body is currently > > > just a binary blob for non-text/* types, otherwise it's text in > > > whatever encoding you've configured. > > > > some kind of common upload target for a single project in order to have > > more reliable sources of stuff thats related to bugs but doesnt fit into > > the normal repository > > Sorry, I'm still having trouble with "doesn't fit into the normal > repository". It's going to be large wherever you keep it. You > worried about multiple branches all having these big tarballs in them > and want a "lightweight" checkout without all the big > tarballs/whatever? I still think having some sort of "resources" > directory on an http server somewhere that you link to from comments > is the best plan. If you use the > be show --xml ID | be-xml-to-mbox | catmutt > approach, you can even write your comments in text/html and get > clickable links ;). A "push big file to remote and commit comment > linking to it" script would be pretty simple and keep everything > consistent. thats probably what i want to do > > > > On Sun, Jul 12, 2009 at 12:57:35AM +1000, Ben Finney wrote: > > > > Ronny Pfannschmidt <Ronny.Pfannschmidt@gmx.de> writes: > > > > > > > > > i want to see the combination of the bug data of all branches > > > > > > > > How is a tool to determine the set of “all branches”? The distributed > > > > VCS model means that set is indeterminate. > > > > > > He could just make a list of branches he likes. > > > > > > Ronny, are you looking to check bug status across several repos on the > > > fly, or periodically run something (with cron, etc.) to update a > > > static multi-repo summary? > > > > on the fly access > > Then listing bugs in a remote repo will either involve httping tons of > tiny values files for each bug (slow?) or running some hypothetical > BE-server locally for each repo speaking some BE-specific protocol > (complicated?). And how would you handle e.g. headless git repos, > where nothing's even checked out? > > You could always run the cron job every 15 minutes, and rely on > whatever VCS you're using having some intelligent protocol/procedure > to keep bandwidth down ;). If you need faster / more-efficient > updates, you'll probably need to throw out polling altogether and > setup all involved repos with a "push to central-repo on commit" hook > or some such. its intended to run on the place where i publish the repositories anyway |
Comment: |
--------- Comment --------- ID: 30a8b841-98ae-41b7-9ef2-6af7cffca8da Short name: bea/12c/30a From: W. Trevor King <wking@drexel.edu> Date: Mon, 13 Jul 2009 06:47:15 -0400 On Mon, Jul 13, 2009 at 09:05:34AM +0200, Ronny Pfannschmidt wrote: > On Sun, 2009-07-12 at 19:55 -0400, W. Trevor King wrote: > > On Sun, Jul 12, 2009 at 11:20:10PM +0200, Ronny Pfannschmidt wrote: > > > On Sat, 2009-07-11 at 11:25 -0400, W. Trevor King wrote: > > > > On Sat, Jul 11, 2009 at 03:13:05PM +0200, Ronny Pfannschmidt wrote: > > > > > On Sat, 2009-07-11 at 08:50 -0400, W. Trevor King wrote: > > > > > > On Sat, Jul 11, 2009 at 01:54:54PM +0200, Ronny Pfannschmidt wrote: > > > > > > > 2. is there any model for storing bigger files at a central place (for > > > > > > > some of my bugs i have multi-megabyte tarballs attached) > > > > > > > > > > > > be comment ID "See the tarball at http://yourpage/something.tar.gz" > > > > > > Then to grab the tarball, you'd use: > > > > > > wget `be show COMMENT-ID | sed -n 's/ *See the tarball at //p'` > > > > > > to grab it. > > > > > > > > > > so the basic idea is to do it completely self-managed > > > > > and have have heterogenous sources of extended data? > > > > > > > > I assume "extended data" here refers to your tarballs. What sort of > > > > homogenous source did you have in mind? The comment body is currently > > > > just a binary blob for non-text/* types, otherwise it's text in > > > > whatever encoding you've configured. > > > > > > some kind of common upload target for a single project in order to have > > > more reliable sources of stuff thats related to bugs but doesnt fit into > > > the normal repository > > > > Sorry, I'm still having trouble with "doesn't fit into the normal > > repository". It's going to be large wherever you keep it. You > > worried about multiple branches all having these big tarballs in them > > and want a "lightweight" checkout without all the big > > tarballs/whatever? I still think having some sort of "resources" > > directory on an http server somewhere that you link to from comments > > is the best plan. If you use the > > be show --xml ID | be-xml-to-mbox | catmutt > > approach, you can even write your comments in text/html and get > > clickable links ;). A "push big file to remote and commit comment > > linking to it" script would be pretty simple and keep everything > > consistent. > > thats probably what i want to do #!/bin/bash REMOTE_DIR="you@webhost:./public_html/bigfiles" REMOTE_LINK="http://www.webhost.com/bigfiles" if [ $# -ne 2 ]; then echo "usage: $0 ID BIGFILE" exit 1 fi ID="$1" BIGFILE="$2" be comment "$ID" "Large file stored at ${REMOTE_LINK}/${BIGFILE}" && scp "$BIGFILE" "${REMOTE_DIR}" > > > > On Sun, Jul 12, 2009 at 12:57:35AM +1000, Ben Finney wrote: > > > > > Ronny Pfannschmidt <Ronny.Pfannschmidt@gmx.de> writes: > > > > > > > > > > > i want to see the combination of the bug data of all branches > > > > > > > > > > How is a tool to determine the set of “all branches”? The distributed > > > > > VCS model means that set is indeterminate. > > > > > > > > He could just make a list of branches he likes. > > > > > > > > Ronny, are you looking to check bug status across several repos on the > > > > fly, or periodically run something (with cron, etc.) to update a > > > > static multi-repo summary? > > > > > > on the fly access > > > > Then listing bugs in a remote repo will either involve httping tons of > > tiny values files for each bug (slow?) or running some hypothetical > > BE-server locally for each repo speaking some BE-specific protocol > > (complicated?). And how would you handle e.g. headless git repos, > > where nothing's even checked out? > > > > You could always run the cron job every 15 minutes, and rely on > > whatever VCS you're using having some intelligent protocol/procedure > > to keep bandwidth down ;). If you need faster / more-efficient > > updates, you'll probably need to throw out polling altogether and > > setup all involved repos with a "push to central-repo on commit" hook > > or some such. > > its intended to run on the place where i publish the repositories anyway Oh, you mean all the repos you want to cover are all _already_ on the same host? -- This email may be signed or encrypted with GPG (http://www.gnupg.org). The GPG signature (if present) will be attached as 'signature.asc'. For more information, see http://en.wikipedia.org/wiki/Pretty_Good_Privacy My public key is at http://www.physics.drexel.edu/~wking/pubkey.txt |
Comment: |
--------- Comment --------- ID: 46937fd4-b0bc-4eed-8033-d699445441ea Short name: bea/12c/469 From: W. Trevor King <wking@drexel.edu> Date: Mon, 13 Jul 2009 07:57:34 -0400 On Sat, Jul 11, 2009 at 11:25:07AM -0400, W. Trevor King wrote: > The easiest implementation I can think of would be to keep local > branches (on whatever computer is hosting your web interface) > following your favorite repos. > proxectX/ > |-- repoA > |-- repoB > `-- repoC > You'd pull upstream changes with a cron job. > Listing bugs would be something along the lines of > projectX$ for repo in * > do > pushd $repo > be list > popd > done | sort | uniq > ... I've reworked option handling for be, so my branch now supports projectX$ for repo in * do be --dir $repo list done | sort | uniq etc. This also makes it easy to use your uninstalled development version of be on any bug directory on your local machine. -- This email may be signed or encrypted with GPG (http://www.gnupg.org). The GPG signature (if present) will be attached as 'signature.asc'. For more information, see http://en.wikipedia.org/wiki/Pretty_Good_Privacy My public key is at http://www.physics.drexel.edu/~wking/pubkey.txt |
Comment: |
--------- Comment --------- ID: c8283e08-967c-4a7b-b953-3ec62c83fb9f Short name: bea/12c/c82 From: Gianluca Montecchi <gian@grys.it> Date: Mon, 13 Jul 2009 10:58:59 +0200 On Sun, Jul 12, 2009 at 12:57:35AM +1000, Ben Finney wrote: > Ronny Pfannschmidt <Ronny.Pfannschmidt@gmx.de> writes: > > > i want to see the combination of the bug data of all branches > > What is your definition of ???all branches???? When I'm working with > distributed VCS, I create branches wherever I feel like, and the VCS > tool doesn't have some central registry of branches to keep up to date. > > How is a tool to determine the set of ???all branches???? The distributed > VCS model means that set is indeterminate. In the first main Ronny spoke about "public" branches. To me it means that if a branch is public, he should like to have a status of that branch. We all agree (probably ;-) ) that tha main branch is the "right" branch, but as I see it, Ronny's question has some logic. I'd like to know that a certain bug is fixed in a certain branch, also if it is still not merged in the main branch, for various reason (ie I am interested in the solution since the bug stop my work) Imagine it like a rss feed aggregator: in one place there are all the bugs of all the branches that the developers make avaible to the public with a repository. This can make easier the life to who want to try a something since he know what branch he must check out, instead of checking all the branch he can find to test if he get what is looking for. Unluckyly I have no idea how to solve it. :-( bye Gianluca _______________________________________________ Be-devel mailing list Be-devel@bugseverywhere.org http://void.printf.net/cgi-bin/mailman/listinfo/be-devel |
Comment: |
--------- Comment --------- ID: 624a4542-92e9-442e-b71c-a14da4fe55cf Short name: bea/12c/624 From: W. Trevor King <wking@drexel.edu> Date: Sat, 05 Dec 2009 22:39:07 +0000 I read http://weblog.masukomi.org/2008/1/3/distributed-bug-tracking yesterday, and the section on bug visibility got me thinking about bug 12c (Multi-repo meta-BE?) some more. We already have interfaces like this email/html mashup: On Sun, Sep 13, 2009 at 07:04:05AM -0400, W. Trevor King wrote: > Since the non-bzr interfaces to BE are coming along nicely, I've put > up a non-bzr interface to my be-rr branch. > http://www.physics.drexel.edu/~wking/code/be > It uses nightly builds of Gianluca's static html from my devel branch > to provide read-only browsing, and accepts changes from the general > public through my email interface into a public branch. I handle the > synchronization of these two branches manually. These interfaces provide a means for remote users to access a BE repository without bzr or the command line. As far as users are concerned, this exposed repository looks pretty much like a centralized bugtracking system (e.g. bugzilla, ...). However, with BE we have more bug information living off in other branches that haven't yet been merged with the exposed repo. The problem is two-fold: 1) how to keep up to date within a distributed community. 2) how do users find branches/patches that fix bug XYZ. For (2), I think the best solution at the moment are along the lines of my little scripts (discussed in the bug 12c comments). With the addition of the `be diff --dir DIR` option, it's now even easier to find more information on bug 565 (or whatever UUID): be/be.wtk$ for repo in ../*; do \ if [ $repo == "be.wtk" ]; then continue; fi; \ diff=$(be diff --dir $repo --subscribe 565:all); \ if [ -n "$diff" ]; then \ echo "Changed from $repo:"; echo "$diff"; \ fi; \ done Changed from ../be.html: New bugs: 565:fm: be email-bugs for bug submission from bzr-less users Changed from ../be.trunk: New bugs: 565:fm: be email-bugs for bug submission from bzr-less users Changed from ../cherryflavoredbugseverywhere: New bugs: 565:fm: be email-bugs for bug submission from bzr-less users where the --dir and --subscribe options to `be diff` are new. If people don't like the command line, this would be easy to bundle into a web-frontend (CFBE?) if you wanted, with a cron job pulling updates into the tracked branches. I was starting into a solution for (1) when I did this: On Mon, Jul 27, 2009 at 08:42:19AM -0400, W. Trevor King wrote: > My email interface now supports subscription: > be subscribe DIR # see any changes to the bug directory. > be subscribe BUG-ID # see changes to a particular bug. > See > be subscribe --help > for more details. The idea was that a dev/user would subscribe to whatever issues they wanted to track, and they would get email notifications whenever some action affected any of those issues. These subscriptions would percolate through the distributed branches as a result of the usual mergers. For example, my subscription to all changes has made it into the trunk branch (see .be/settings). This subscription mechanism was setup to work through interactive public interfaces (my email interface, eventually CFBE, ...), but it doesn't work for changes made via the command-line interface, so I browsed around a bit and ran across some interesting workflows in the bzr documentation doc/developers/HACKING.txt, "Communicating and Coordinating" which points out the following plugins * email (http://doc.bazaar-vcs.org/plugins/en/email-plugin.html) * dbus (http://doc.bazaar-vcs.org/plugins/en/dbus-plugin.html) which send automatic notification messages after commits, etc. If people want this sort of functionality, it would be easy enough to rig a hook for `be commit' that sent a diff email to subscribers, which could include be-devel. |
ID: 88d1f2c2-e1af-4f0d-9390-e3c89ae4f7d7
Short name: bea/12c/88d
From: Ronny Pfannschmidt <Ronny.Pfannschmidt@gmx.de>
Date: Sat, 11 Jul 2009 13:54:54 +0200