1 2010-09-18T00:35:36  *** RogerHaase has left #moin-dev
   2 2010-09-18T01:43:04  *** xorAxAx has quit IRC
   3 2010-09-18T01:43:50  *** xorAxAx has joined #moin-dev
   4 2010-09-18T02:45:06  *** xorAxAx has quit IRC
   5 2010-09-18T02:45:52  *** xorAxAx has joined #moin-dev
   6 2010-09-18T03:17:44  *** grzywacz has quit IRC
   7 2010-09-18T05:45:11  *** valeuf has joined #moin-dev
   8 2010-09-18T10:42:01  <dreimark> moin
   9 2010-09-18T11:32:09  <ThomasWaldmann> moin
  10 2010-09-18T11:36:23  <dreimark> moin
  11 2010-09-18T13:10:20  <ThomasWaldmann> btw, i just did a quick try using flask-cache for dom-tree - that made it quite faster :)
  12 2010-09-18T13:12:29  <ThomasWaldmann> (but i didn't commit it, it was caching the complete internal_representation, which is a bit too much)
  13 2010-09-18T13:21:37  *** grzywacz has joined #moin-dev
  14 2010-09-18T13:21:37  *** grzywacz has joined #moin-dev
  15 2010-09-18T13:50:40  *** grzywacz has quit IRC
  16 2010-09-18T14:47:34  <valeuf>  hey ThomasWaldmann
  17 2010-09-18T14:47:55  <valeuf> long time no see ^^ I have been pretty busy last days, but I am back in the business
  18 2010-09-18T14:49:27  <valeuf> I understood the problem with the bug, I wrote a quick fix, however it is not really clean
  19 2010-09-18T14:49:28  <dreimark> wb valeuf
  20 2010-09-18T14:49:37  <valeuf> moin dreimark
  21 2010-09-18T14:49:51  <valeuf> I am trying to see what I missed in the dev of moinmoin last weeks
  22 2010-09-18T14:52:25  * dreimark much :)
  23 2010-09-18T14:54:39  <valeuf> I am happy to see that things are going well, I will try to get back on the path
  24 2010-09-18T14:57:14  <xorAxAx> valeuf: i havent submitted the linkedin thingie yet, but its still on my todo list
  25 2010-09-18T14:57:46  <dreimark> hi xorAxAx
  26 2010-09-18T14:58:10  <xorAxAx> hi dreimark
  27 2010-09-18T14:58:23  <ThomasWaldmann> moin everybody :)
  28 2010-09-18T14:59:06  <valeuf> xorAxAx: oh thanks :) me too I have a long todo list, so I understand is not your top priority
  29 2010-09-18T14:59:41  <valeuf> ei-grad: diofeher did you receive mail from Facebook recruitement team after your GSoC?
  30 2010-09-18T15:00:55  <ThomasWaldmann> hehe
  31 2010-09-18T15:07:54  <dreimark> bbl
  32 2010-09-18T16:20:44  *** RogerHaase has joined #moin-dev
  33 2010-09-18T16:47:59  *** RogerHaase1 has joined #moin-dev
  34 2010-09-18T16:50:35  *** RogerHaase has quit IRC
  35 2010-09-18T17:25:14  *** RogerHaase1 has left #moin-dev
  36 2010-09-18T18:45:40  <ThomasWaldmann> hmm, we maybe have a problem with cache keys
  37 2010-09-18T18:46:51  <ThomasWaldmann> moin uses (used) a trick to avoid doing acl checks for stuff sent from the cache
  38 2010-09-18T18:47:43  <ThomasWaldmann> it generates the cache access keys involving a server secret, so the user can't compute an access key without the help from server
  39 2010-09-18T18:48:14  <ThomasWaldmann> (this is what one sees as from_cache=... value in the url for such cases)
  40 2010-09-18T18:49:24  <ThomasWaldmann> but caches (like the stuff from werkzeug, flask-cache) don't assure that stuff is available, not even for some specific time
  41 2010-09-18T18:49:43  <ThomasWaldmann> an object can basically be gone any time and the cache returns None then
  42 2010-09-18T18:50:35  <ThomasWaldmann> but if that cache key was all we have, there's no way to re-compute the content
  43 2010-09-18T18:51:58  <ThomasWaldmann> that can be solved by giving all params needed to compute the content
  44 2010-09-18T18:52:57  <ThomasWaldmann> but if that works in some obvious way that can be done by user, he basically has an access method for that content and we'ld need to check acls
  45 2010-09-18T18:53:35  <ThomasWaldmann> any other idea than using symmetric encryption?
  46 2010-09-18T18:59:18  <dreimark> re
  47 2010-09-18T19:03:54  <ThomasWaldmann> (if the method is too expensive, we might be better off with just checking acls)
  48 2010-09-18T19:04:08  <dreimark> not yet, but i am also not sure if encryption is enough or how to do it safe enough
  49 2010-09-18T19:04:41  <dreimark> no cache - no problems
  50 2010-09-18T19:05:25  <ThomasWaldmann> :) yes, but maybe not what one wants for thumbnails and similar stuff
  51 2010-09-18T19:06:28  <dreimark> maint cleancache and recreating is neccessary if the secret becomes changed
  52 2010-09-18T19:07:15  <dreimark> how slow is an acl check?
  53 2010-09-18T19:07:49  <ThomasWaldmann> i have no specific numbers, but slower than doing none :)
  54 2010-09-18T19:08:15  <dreimark> we may be could also use the tar item
  55 2010-09-18T19:08:37  <ThomasWaldmann> needs opening of item metadata (or later maybe getting that from index) and computing may.read for that user
  56 2010-09-18T19:09:13  <dreimark> e.g .if  a tar item has the image and a thumbnail etc. then we need to check only once acls
  57 2010-09-18T19:09:31  <dreimark> and the whole container can be accessed
  58 2010-09-18T19:10:02  <ThomasWaldmann> yes, but at next request, you need to check again (and we are talking about multiple requests)
  59 2010-09-18T19:10:34  <dreimark> true. but we need to check access for the item anyway
  60 2010-09-18T19:10:51  <ThomasWaldmann> well, I guess there is no real choice. :|
  61 2010-09-18T19:11:21  <ThomasWaldmann> i guess i have to implement acl check and do caching the more usual way.
  62 2010-09-18T19:13:00  <ThomasWaldmann> waldi: TheSheep: any other ideas before i do that?
  63 2010-09-18T19:13:27  <dreimark> i think too it needs to be checked on item level. may be we can cache somehow the result if many items are used from one item
  64 2010-09-18T19:13:52  <ThomasWaldmann> that is just asking for more trouble X)
  65 2010-09-18T19:14:28  <ThomasWaldmann> but maybe index with cache would be nice
  66 2010-09-18T19:14:28  <dreimark> probably yes
  67 2010-09-18T19:14:54  <ThomasWaldmann> ehrm, index with acls
  68 2010-09-18T19:16:25  <dreimark> that would make it possible to show what a user can do or has access to
  69 2010-09-18T19:16:55  <dreimark> faster
  70 2010-09-18T19:36:15  *** grzywacz has joined #moin-dev
  71 2010-09-18T21:17:29  *** RaphaelBossek has joined #moin-dev
  72 2010-09-18T21:18:23  <RaphaelBossek> hi folks; I've cloned 2.0-dev looking for the interface where to integrate a possible formater for PDF
  73 2010-09-18T21:21:32  <dreimark> ThomasWaldmann: valeuf ^
  74 2010-09-18T21:24:54  <dreimark> RaphaelBossek: in principle you need to write a convertor pdf_out
  75 2010-09-18T21:25:12  <dreimark> see converter2
  76 2010-09-18T21:26:09  <RaphaelBossek> I've started MoinMoin at localhost:8080 and would like to try some of those convertes out, but how ?
  77 2010-09-18T21:27:38  <dreimark> create an item by appending an item name to the url
  78 2010-09-18T21:28:23  <RaphaelBossek> could you please continue this link if I want DOCBOOK ? http://localhost:8080/Home
  79 2010-09-18T21:30:54  <dreimark> see here http://moinmo.in/DOM%20DocBook%20and%20HTML%202010/DocBook%20Converter
  80 2010-09-18T21:31:10  <ThomasWaldmann> hi RaphaelBossek
  81 2010-09-18T21:31:20  <RaphaelBossek> hi ThomasWaldmann
  82 2010-09-18T21:31:39  * RaphaelBossek have to watch Schlag den Rab ;-)
  83 2010-09-18T21:31:50  <ThomasWaldmann> :)
  84 2010-09-18T21:32:58  <RaphaelBossek> dreimark, I would like to convert the http://localhost:8080/Home page to DOCBOOK in order to go through the code with my debuger
  85 2010-09-18T21:33:21  <dreimark> i know but i can't help much just pointing on documentations
  86 2010-09-18T21:33:36  <dreimark> valeuf: would be the best choice for QA
  87 2010-09-18T21:33:45  <dreimark> ThomasWaldmann: may be able to help too
  88 2010-09-18T21:33:46  <RaphaelBossek> dreimark, it would be helpful to have an action menu item to select the converter2 plugins
  89 2010-09-18T21:34:43  <RaphaelBossek> ThomasWaldmann, try to investigate the possibilities to integrate a PDF converter2 plugin using http://www.reportlab.com/software/opensource/rl-toolkit/download/ for MoinMoin 2.0
  90 2010-09-18T21:35:21  * RaphaelBossek is away
  91 2010-09-18T21:37:14  <ThomasWaldmann> .../+convert/ItemName?mimetype=application/docbook+xml
  92 2010-09-18T21:39:17  <dreimark> do we have a page where all these commands listed?
  93 2010-09-18T21:40:13  <ThomasWaldmann> valeuf: ^^?
  94 2010-09-18T21:40:27  <dreimark> and easy to find please
  95 2010-09-18T22:07:14  <dreimark> ThomasWaldmann: I want all meta http://paste.pocoo.org/show/263689/ (AclWrapperItem)
  96 2010-09-18T22:08:16  <dreimark> for updating usergroup and not killing other meta data
  97 2010-09-18T22:13:28  <dreimark> just wondering how i can access is without 5 lines
  98 2010-09-18T22:16:21  <ThomasWaldmann> what do you intend to achieve?
  99 2010-09-18T22:17:17  <dreimark> http://paste.pocoo.org/show/263696/ just want to have something which I can use for appends
 100 2010-09-18T22:17:33  <dreimark> and not only for USERGROUP
 101 2010-09-18T22:19:40  <ThomasWaldmann> dict(rev) should give a copy of all meta
 102 2010-09-18T22:20:50  <dreimark> cool
 103 2010-09-18T22:30:47  <CIA-36> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 9970:f76734201585 2.0-dev/MoinMoin/support/flaskext/cache/ (__init__.py backends.py): add flask-cache 0.3.2
 104 2010-09-18T22:31:05  <CIA-36> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 9971:0b6524f36604 2.0-dev/MoinMoin/__init__.py: have a Cache instance (see flask-cache) at app.cache
 105 2010-09-18T22:31:05  <CIA-36> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 9972:022553365c74 2.0-dev/MoinMoin/ (caching.py items/__init__.py wikiutil.py items/sendcache.py): (log message trimmed)
 106 2010-09-18T22:31:05  <CIA-36> kill sendcache module (use app.cache directly), move leftovers to caching.cache_key and wikiutil.file_headers
 107 2010-09-18T22:31:05  <CIA-36> made cache_key a bit more generic: just give anything you like to be involved
 108 2010-09-18T22:31:06  <CIA-36> in computation of the cache key as kw args.
 109 2010-09-18T22:31:06  <CIA-36> some optimizations with file-like objects had to be removed, because the cache
 110 2010-09-18T22:31:09  <CIA-36> backends of flask-cache rather want strings, not open files.
 111 2010-09-18T22:31:10  <CIA-36> TODO: transformed/diffed images get put into cache and it is expected that they
 112 2010-09-18T22:31:10  <CIA-36> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 9973:9adc5a29769d 2.0-dev/MoinMoin/ (theme/__init__.py action/pollsistersites.py): use app.cache for sisteritems
 113 2010-09-18T22:31:10  <CIA-36> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 9974:1c6e8694ce18 2.0-dev/MoinMoin/items/__init__.py: TarContainer filling: use temp files, not cache
 114 2010-09-18T22:31:10  <CIA-36> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 9975:f812183014a6 2.0-dev/MoinMoin/caching.py: killed old code from MoinMoin.caching module (not used any more)
 115 2010-09-18T22:31:10  <CIA-36> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 9976:d70544bff831 2.0-dev/MoinMoin/ (apps/frontend/views.py items/__init__.py): (log message trimmed)
 116 2010-09-18T22:31:11  <CIA-36> change caching for image diff and image resizing, details below
 117 2010-09-18T22:31:11  <CIA-36> Old method: compute stuff (diffed/resized image), put into cache, use cache key
 118 2010-09-18T22:31:36  <CIA-36> that is computed involving a server secret. Then access the cache using this
 119 2010-09-18T22:31:36  <CIA-36> cache key. This relies on that the cache still has the data (which might not be
 120 2010-09-18T22:31:36  <CIA-36> the case). Cache key must not be guessable, no ACL check needed to read the
 121 2010-09-18T22:31:36  <CIA-36> resized/diffed image (if a user got the cache key, the user has the required
 122 2010-09-18T22:31:36  <CIA-36> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 9977:f8ab3f94bc55 2.0-dev/MoinMoin/ (11 files in 6 dirs): simplify cache_key, kill caching module, move cache_key to wikiutil
 123 2010-09-18T22:31:36  <CIA-36> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 9978:236e8f9c6d4c 2.0-dev/MoinMoin/ (items/__init__.py theme/__init__.py): add usage string to cache_key() params to avoid collisions
 124 2010-09-18T22:40:10  <CIA-36> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 9979:a5f42633e37e 2.0-dev/wikiconfig.py: add configuration example for flask-cache
 125 2010-09-18T22:45:33  *** grzywacz has quit IRC
 126 2010-09-18T23:18:54  *** RaphaelBossek has quit IRC
 127 2010-09-18T23:28:52  <CIA-36> Thomas Waldmann <tw AT waldmann-edv DOT de> default * 9980:c9cf24017419 2.0-dev/ (4 files in 4 dirs): get rid of cache_dir, action/cache secret

MoinMoin: MoinMoinChat/Logs/moin-dev/2010-09-18 (last edited 2010-09-17 22:45:03 by IrcLogImporter)