This may be useful: http://www.ke-cai.net/2010/05/tracking-change-with-google-diff-match.html http://neil.fraser.name/software/diff_match_patch/svn/trunk/demos/demo_diff.html
look at the page source. It uses: http://neil.fraser.name/software/diff_match_patch/svn/trunk/javascript/diff_match_patch.js On Oct 24, 11:15 pm, mart <msenecal...@gmail.com> wrote: > k, items 1 and 2 should be simple enough.... item 3, that is a > different story, but ok, sure. I think merging outside of the > mercurial context is best though, then simply update the remote > repository with the merged file (and include good notes describing > both source files and capturing the diffs of the merge is best). > > I will be under the gun at work for the next few days, but after that > will be able to have something for you - by end of week is a good > buffer. > > 1 question: how large of added disk space usage would you is > acceptable - I have an idea that I will look into. (I always favor > having less network IO, and more disk IO when possible). > > Let me know if end of week is good for you. > > Mart > > On Oct 24, 11:26 pm, mdipierro <mdipie...@cs.depaul.edu> wrote: > > > I would be happier with much less... > > > 1) a button in the /admin/mercurial/revision/<n> page that says [diff] > > and provides a text output with a diff between the revision and the > > current code > > > 2) a few more fields in the /admin/mercurial/commit page that say > > > changelog: write something here > > repository: https://<....>.googlecode.com/hg/ > > push: yes/[no] (if yes code is pushed in googlecode) > > pop: yes/[no] (if yes, current code is cloned from googlecode and > > overwritten) > > > I think of [merge] as a third order approximation and it is difficult > > to get it right. > > > Massimo > > > On Oct 24, 10:16 pm, mart <msenecal...@gmail.com> wrote: > > > > K, good stuff! I'll use what you and Boris came up with and integrate > > > to mine (will help speed things up) the other project :) and for the > > > other thing, great idea (i think). what do you mean by "web > > > repository"? do you mean reproduce from the code behind a live web2py > > > server instance? which would be a grand idea! if that's what you mean. > > > I would doable if a well kept manifest (generated through automation > > > on "pull") describing in xml format file names (path) and rev # be > > > installed long with web2py. then a simple xmlrpc meg over https should > > > get you a nice xml file wich can be used to resolve the variables used > > > in a hg pull cmd. (at least that's what I'd do to start) - I do > > > something like that to get build automation to talk to the bug > > > tracking system (keeps QA happy with reports like "build X contains > > > THIS list of bug fixes as described in THIS list of changelist > > > description, which contains THIS list of affected files, which was > > > checked in by THIS user, and reviewd by etc...." > > > > But, please let me know if I got the requirement all wrong, and I'll > > > be happy to suggest another approach. > > > > For the diff requirement: I would definitely start fro the bottom up. > > > This is what I do currently do with Perforce (logic should be > > > similar). > > > > 1) I generate an XML representation of a directory structure (where I > > > need to be able to specify any folder within my dir structure as the > > > "root" folder (the starting point), and need to be able to exclude > > > files and/folders as params to the function call. > > > > 2) once I have my xml object (I like a well structured Element Tree > > > for this type of thing), I will simply pass the elements (in a loop) > > > to a function that will have the know-how to return a correct mapping > > > between repository and workspace (i think Mercurial calls this the > > > Working Directory). > > > > 3) diff the file, capture the diff (if any) store it (keeping the the > > > return strings in order is important, so I keep each return values as > > > a dictionary within warm brackets of a collection that respects order > > > (like a deque()). > > > > once done and have gone through the the complete fileset, then it just > > > depends on how fancy we want to be (color, special indentation, format > > > to 2 pages, etc..._) or the complete returned values can be formatted > > > and handed to araxis (or something like that)... > > > > in the .hg word, its a little more convoluted in that I believe > > > another repository needs to be created and changes pulled from the > > > from the revision you are interested in (can be any revision or > > > "latest" but it does need to be specified. Would look something like > > > this: > > > > mkdir myTempDir > > > cd myTempDir > > > hg init > > > hg pull myTempDir > > > hg update -r N > > > > then do that recursive diff as discussed above > > > > is this what you were looking for? > > > > Mart :) > > > > connect to app admin through script: > > > On Oct 24, 9:14 pm, mdipierro <mdipie...@cs.depaul.edu> wrote: > > > > > Actually I just did this with some help from Boris from the mercurial > > > > mailing list. > > > > > Now if you run web2py from source ad you have mercurial installed > > > > (easy_install mercurial) you can use the web2py admin to: > > > > > - create a repository (this is done automatically and it also makes > > > > a .hgignore) > > > > - commit file > > > > - see log changes > > > > - see list of files for each committed revision > > > > - revert to past revision > > > > > (all by just clicking) > > > > > One important missing feature is the ability to get a diff vs a given > > > > revision. > > > > Another feature I would like to have is the ability to push from a web > > > > repository or revert to it. > > > > > On Oct 24, 8:09 pm, mart <msenecal...@gmail.com> wrote: > > > > > > Hi Massimo, > > > > > > Ok, now I feel bad... I have not had the time to finish the > > > > > integration app I promised... Much is done, but not yet complete... I > > > > > still do intend on completing though... hopefully soon... > > > > > > Until then, what is your intent here? are you looking to restore a > > > > > fileset (or entire tree) to a specific version from repo (without > > > > > changing meta data), are you looking to drop all meta changes, revert > > > > > you local code line to known state? > > > > > > are you looking for a quick and dirty recipe like with combination of > > > > > cmds to delete local changesets and revert to a speciic version (or > > > > > #head revision) of the remote depot? like: > > > > > delete local repo, then: > > > > > hg init --> create new > > > > > hg pull --> get a version > > > > > hg update --> well, just update... > > > > > > Although, by reading your commit script, I'd say you're pretty handy > > > > > with mercurial API as it is. > > > > > > anyways, depends what you are looking to do... I can give you some > > > > > quick and dirty recipes if you like, if it helps. > > > > > > Mart :) > > > > > > On Oct 24, 6:23 pm, Massimo Di Pierro <mdipie...@cs.depaul.edu> wrote: > > > > > > > Hello everybody, > > > > > > > I am working on improving the web2py web interface to mercurial. > > > > > > >http://127.0.0.1:8000/admin/mercurial/commit > > > > > > > We can current create a repo, commit and get changelog but I cannot > > > > > > figure out how to do revert. > > > > > > Here is the code I have: > > > > > > > import os > > > > > > uio = ui.ui() > > > > > > uio.quiet = True > > > > > > if not os.environ.get('HGUSER') and not uio.config("ui", > > > > > > "username"): > > > > > > os.environ['HGUSER'] = 'web...@localhost' > > > > > > try: > > > > > > repo = hg.repository(ui=uio, path=path) > > > > > > except: > > > > > > repo = hg.repository(ui=uio, path=path, create=True) > > > > > > > given repo and a revision number from repo.changelog, how to I > > > > > > revert all files to that revision using the API? > > > > > > > Massimo > >
