Bugs item #3502940, was opened at 2012-03-12 15:00 Message generated for change (Settings changed) made by kargor You can respond by visiting: https://sourceforge.net/tracker/?func=detail&atid=843359&aid=3502940&group_id=167540
Please note that this message will contain a full copy of the comment thread, including the initial issue submission, for this request, not just the latest update. Category: Performance Group: None Status: Closed Resolution: Fixed Priority: 5 Private: No Submitted By: Stefan Rossbach (kargor) >Assigned to: Stefan Rossbach (kargor) Summary: "Scan Workspace" creates expensive file lists Initial Comment: This feature scans your current Eclipse workspace for a project that has the best match to the project that the inviter is send to you. Problem: the computeMatch computes a match based on the filenames. To compute the match, you need 2 file lists. When these file lists are created, the checksum calculation will be performed automatically. Scenario: My workspace constists of 10 projects, total size 2 GB and about 50000 files. If I press this button ... well, you can go and make you a cup of coffee. At least if you have a HDD and not a SSD. ---------------------------------------------------------------------- Comment By: Stefan Rossbach (kargor) Date: 2012-11-13 09:39 Message: http://dpp.git.sourceforge.net/git/gitweb.cgi?p=dpp/saros;a=commit;h=26f2ff7190f0eb1e693f2c7736ef6abdd80dc6d4 ---------------------------------------------------------------------- Comment By: Stefan Rossbach (kargor) Date: 2012-07-23 00:50 Message: Feature was removed. ---------------------------------------------------------------------- Comment By: Stefan Rossbach (kargor) Date: 2012-06-28 15:01 Message: We should add more algorithms to Saros with a complexity of O(c^n). At least performance related problems will no longer be downgraded too stupid priority numbers ... if the application is too slow, get better hardware ... http://thedailywtf.com/Articles/Throw-Some-Hardware-at-it!.aspx ---------------------------------------------------------------------- Comment By: Alexander Waldmann (netcorps) Date: 2012-06-28 09:26 Message: There are numerous features that were added to saros without proper requirements management. It's unsurprisingly hard to get that junk out of the software, because the usual developer is hesitant in removing something... after all feature count is great, right...? This feature is useless. Remove it already, as it serves no purpose. Still discussing it over and over is even worse. This bug is probably worded in a wrong way - it should say "Remove useless, time consuming feature". Scanning does NOT do what the user expects, if there are multiple projects, as this just stops at the first match of over 80%. So if you want to use it to find the BEST matching project, it will NOT do that, unless it accidentally tries the best match first while scanning. If someone has a workspace and NO idea which project is incoming right now, he/she is in serious trouble, and this feature won't help anyway, so for finding "the one matching project" this is way too slow and useless, too. If someone clicks on the button, dataloss is possible as the computer might even freeze on old machines. There is no proper progress indication, so whenever that button is pressed, the user is guarantueed to have a bad experience. ---------------------------------------------------------------------- Comment By: Franz Zieris (franzzieris) Date: 2012-06-28 03:41 Message: > Maybe you should read the problem again I did :) We are talking about (at least) two different subjects: (1) The actual performance, as stated in this bug's category and (2) the usability of the system ("Keep the GUI responsive.") The explanation in my previous comment was meant to justify the lowering of the priority from "8" to "5". You can read it as: "Well, I used a pretty big workspace on an average laptop and all the program did was rattling for half a minute. Nothing bad happened, the result was fine. It just took a while. Thus level 8 on a scale ranging from 1 to 9 seems quite exaggerated." Nothing more, nothing less. It does not matter how fast this process really is and what technical dodges to increase some throughput which system is capable of. ---------------------------------------------------------------------- Comment By: Stefan Rossbach (kargor) Date: 2012-06-27 12:38 Message: Maybe you should read the problem again, we calculate something that we do not need for the result. As for your results. Congratulation, 80 MB/s Average read throughput on my HDD 20 - 30 MB / s (very old IBM HDD) Average read throughput on my notebook HDD 10 MB /s Not taken into account, that your OS may have cached most of the files already into memory, the physical placement of the files on the disk which can reduce througput. And no, I would not expect that my GUI stalls for minutes when I press a button. Keep the GUI responsive. ---------------------------------------------------------------------- Comment By: Franz Zieris (franzzieris) Date: 2012-05-23 08:59 Message: Since this only takes some time and doesn't actually harm anyone, I lowered the priority from 8 to 5. I just tested the master branch (Git 0b7b758) with a similar workspace (5 projects, 2 GB, 20000 files) and the scan took about 25 seconds. If somebody really uses a workspace that big he/she will understand that scanning this mammoth will take some time. ---------------------------------------------------------------------- Comment By: Stefan Rossbach (kargor) Date: 2012-03-13 15:57 Message: Please remove this feature and close this bug, ---------------------------------------------------------------------- Comment By: Alexander Waldmann (netcorps) Date: 2012-03-13 11:48 Message: This feature can be removed which i did in my prototype - no one missed it, no one used it more than once in the usability tests anyway. It is too slow to work for the average user. It isn't of help - manually choosing the locally existing project is much faster (user knows in most cases why he renamed it). Quickscan = compare projectname works in most cases. If there are two almost identical projects this yields undesired results: the first project is taken that matches by at least 80%, no further scanning is done, so if there is a project that appears in a later position in my workspace, it is not found. One of the useless functions that were introduced because (probably) ONE user asked for it, without thinking through if this actually helps (even that one user) or is yet another thing most users will never user / get confused by / ignore.. ---------------------------------------------------------------------- You can respond by visiting: https://sourceforge.net/tracker/?func=detail&atid=843359&aid=3502940&group_id=167540 ------------------------------------------------------------------------------ Monitor your physical, virtual and cloud infrastructure from a single web console. Get in-depth insight into apps, servers, databases, vmware, SAP, cloud infrastructure, etc. Download 30-day Free Trial. Pricing starts from $795 for 25 servers or applications! http://p.sf.net/sfu/zoho_dev2dev_nov _______________________________________________ Dpp-robot mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/dpp-robot
