Hi,
can you give me help to get mysql running as easy as possible on debian. Then I can try help you out. As far as ODBC with MySql goes I only have the very old doc by Paolo Marchesi (appendix 2?). I prefer to skip odbc as far as possible. I will make the mysql download as suggested.
Marcel
For NATIVE mysql support (ODBC not used):
WINDOWS: You need to use the mysql client dll of 5.022 (that what's dgate and the GUI are compiled against) even if you are using a newer mysql server . I wanted to redistribute the file but got no clear reply from the MySql representative that it was allowed. The PDF is in dicomserver1412.zip
LINUX: your attempts look good - only I think DoubleBackslashtoDB should be 1 (minor issue). Don't know what is the problem opening the database. Any clue in the compile time messages?
Marcel
Just tested the US sequence with image 11 a cine. Works fine (uncompressed or NKI compressed) in conquest14.12 with kpacs0.9.5.1. Probably the viewers don't like series with mixed stills and cine and mixed RGB with MONO
Marcel
Jpeg should be DICOM compliant. Recent testing showed some bugs that are currently being fixed though.
The only manual is dgate -?. This listing is also printed in conquestpacs.pdf. For more info you will have to look at the implementation in dgate.cpp in dgate1412.zip starting at line 8386. 
Here commands sent from another 'dgate --xxxx" are processed. Around line 3585 the older commands 'dgate -x' are processed.
Marcel
Hi,
the first chapters of the manual is its conformance statement. Conquest is DICOM3 compliant - but details about what that means are in teh manual
Marcel
cache is r/w
jukebox is read only copy,
so at one time data will be the same
after verify cache will be deleted
Seems like the only safe way to burn a DVD to me.
I do not understand the other issue? We burn on a timer using a separate app that runs when disk space is low.
Marcel
NKI compression is lossless!
No,
the server maintains a single link to the data, so only one copy exists. You can mirror MAG0 or copy jukebox dirs by hand.
Marcel
Hi BookWorm,
I am suprised that the hard path was a problem - that is only there for WIN32. Please mail me the changed file. Are you using 1.4.12 and mysql 5.0.22? The connect code is in odbci.cpp, have a look there.
Marcel
Probably edting odbci.hpp is enough. Did you create an empty database.
The cgi interface is independent of that database. It will not work though before the server is up.
QuoteIt does not work, if these dicom files are 1. compressed like nki-compression
We have not seen this issue. Can you provide me with full details?
Marcel
Files dropped into the conquest server will be copied to the first MAG device with enough space. Drag and drop behaves the same as a network transfer into the server. There is no way to control in which MAG device images are stored.
Marcel
Since there has been some discussion about the validity of the private NKI compression (see topic Conquest and Ultrasound), here are my views.
NKI compression is an old and well tested feature of the Conquest DICOM server. We use it to store TB's of information. The compression ratio and speed are quite OK.
It is totally transparant for clients accessing conquest over the network, guaranteed error free, and safe to use.
Also (when set in acrnema.map) it can set as transfer syntax to transfer data from one server to another. This is OK for Conquest to Conquest transfers but should NEVER be done for transfers from Conquest to non-Conquest clients.
The compressed data is saved in a private TAG if possible. If not, the images are not compressed and remain valid.
Files stored by Conquest under NKI compression are NOT DICOM COMPLIANT and are unreadible for other software. Originally, it was not possible to store NKI compressed images with the .dcm extension (set with FileNameSyntax), but only with the .v2 extension (raw VR dump). That the combination of .dcm and NKI compression is now possible can actually be considered a bug - .dcm files must be DICOM compliant. Since version 1.4.12b, the nki compression is disabled for .dcm files.
If you store data with NKI compression, you will need to keep the a version of the Conquest software running indefinitively to be able to read the data (or reuse our code in nkiqrsop.cxx).
As David Clunie mentioned, a better way to implement NKI compression is a private transfer syntax. We will consider this change, but it has a lot of implications for the code.
Marcel
The dicom.ini file does not allow # comments on used lines, only on empty ones. The password and username will not be correct. Also, what did you change to odbci.cpp?
Marcel
Hi,
thanks, I will take your changes as much as possible. MySql under linux was never implemented (as documented), but should not but hard to do.
Marcel
Change requests for next release after 1.4.12c:
1. Private transfer syntax for NKI compression. Under consideration.
2. Create AVI's through the built-in k-pacs viewer. Under consideration.
3. Extensions to web interface (user contribution, see postings) + K-pacs based web viewer using cgi and apache or iis (see simple web viewer). I have now prototyped a ActiveX viewer and added it to the cgi interface. A question is if the dicom files should be served up directly from storage (fast, unflexible, harder to setup) or through the cgi interface (slow but flexible, e.g., images can be downsized and compressed). In 1.4.13alpha.
4. Include noob manual for WAMP (user contribution). In 1.4.13beta.
5. Automatically create sql, mysql, or postgres databases during install. In 1.4.13beta.
6. Add 'compression xx' import converter to set preferred compression. In 1.4.13beta. Also allow generate new UIDs like when changing patient id. Under consideration.
7. Transfer entire patient when forwarding some images (prefetch, preforward). Under consideration. In 1.4.13beta
8. Allow permanent debug logging through dicom.ini. Todo.
9. Allow named pipes for debuglog_on and log_on commands. In 1.4.13alpha.
10. BB: Darwin port of dgate - source code mods. In 1.4.13beta.
11. Maybe it would be possible to added checkboxes in the web interface series list and allow opening one or more series of one study to separate viewers.
12. urologic: I'll think of a way to defer forwards. In 1.4.13alpha.
13. Allow overriding return port for C-MOVE using e.g. CONQUESTSRV1~p1 as called AE. In 1.4.13beta
14. Current selection for deletion and move of patients is accesstime of first store in database. It should maybe be changed into patients sorted on latest studydate. In final release 1.4.13.
15. Set max number of failed retries. In final release 1.4.13.
16. 5 characters in port edit box. In 1.4.13beta
Marcel
Hi Ted,
I condensed your code for the forum. I will include it in the next build. However, what does the centernumber mean?
Marcel
Here is a list of known issues in release 1.4.12c. This list will be regularly updated.
2005: Secure socket layer error with GUI browser (+/-BDE) on some installations of MSserver2003 with MSSQL2005. Server runs fine. Unresolved, but can be circumvented using the BrowseThroughDBF flag in upcoming final release 1.4.13.
20061231: 'forward' exportconverters now just hangup the connection when done rather than doing a proper DICOM release due to a problem in PDU.CXX. The behavior is changeable from 1.4.12b to experiment (see ForwardAssociationRelease). Do not know if this is a real issue yet.
20070127: Lambert is working on a crash in the jpeg decompressor. Fixed in final release 1.4.13.
20070127: The problem of image corruption of ACUSON US movies with jpeg compression is unresolved. I found one problem when regenerating a dbase with these images stored as .v2: they would not load because of a very large header (maxheadersize was too small, fixed in 1.4.13alpha).
20070203: Under linux with the built-in dbaseIII driver (tested on an ext3 partition), the image database can grow no larger as 2 GB. This means up to 1.486.146 images can be stored. Attempts to store more images FAIL WITHOUT ANY ERROR MESSAGE. This will also be true for previous versions. This problem does not occur under windows, I have a sample database of 6GB big (4.600.000 images). Fixed in 1.4.13alpha.
20070204: The GUI does not generate an unique UIDPrefix at install(already not in 1.4.9). This means that all conquest dicom servers use the same default UID prefix when renaming, splitting, merging or anonymizing dicom studies. There is therefore a (very) slight change that files generated in different servers receive the same UID. Fix for now by manually adding "UIDPrefix = 1.2.826.0.1.3680043.2.135.xxxxxx.xxxxxxx" where x is a number to dicom.ini. Fixed in 1.4.13alpha.
20070206: The manual does not show %studydate and e.g., %studydate[0,3] as options to get date or a substring of the date (this one gives the year) in the flexible filenamesyntax. Fixed in 1.4.13alpha.
20070207: Worklistmode processing does not work correctly. The string where the part after 'where' is collected is not cleared. Must have worked during testing by accident. Fixed in 1.4.13alpha.
20070207: Only in DbaseIII mode, invalid images with same UIDs but different patientID, StudyUID or SeriesUIS are accepted while they should be rejected. Will be fixed at study level in 1.4.13alpha. Unresolved at lower levels because of the severe performance impact.
20070211: Web interface fails for patient ID with spaces or '. Unresolved.
20070211: The normal install did not create indices for mysql. To create the indices manually for now, regenerate with 'dgate -v -r'. Fixed in 1.4.13alpha.
20070213: Sorting of images list in web interface (imagelister task) is incorrect. Fixed in 1.4.13alpha for seriesviewer, unresolved for image list.
20070213: decompression of mode n3 sometimes fails in K-PACS viewer (p20401977:47). Fixed in 1.4.13alpha (bug in K-PACS viewer).
20070213: table names in cgi interface are passed as lowercase; should have correct case for mysql under linux (rexford.newbould), i.e., need to read them from dicom.ini. Fixed partly (not read from dicom.ini but use correct case) in 1.4.13alpha.
20070215: when no association is accepted there is insufficient logging to illustrate why the server refuses a connection. It just shows 'thread ended'. Fixed in 1.4.13alpha.
20070226: mp: Merging >15 series at once crashes the server due to uids[] array overrun. Fixed in 1.4.13alpha.
20070226: mp: Dicom.ini reads from current dir but not dgate dir. In 1.4.13beta.
20070226: mp: Dicom.ini does not allow tabs around the =. In 1.4.13beta
20070302: scoots39: manual error %studyid and %seriesid. Fixed in 1.4.13alpha. Drag and drop of files with , in filename fails. In 1.4.13beta.
20070305: Indeed stationname should be in the series table. Can someone check the location of all fields in the database against the dicom standard? TODO.
20070315: C_STORE requests due to a move does not contain items 0000,1030 (MoveOriginatorAE) and 0000,1031 (MoveOriginatorMessageID). This caused a kodak web viewer to fail when accessing a conquest server. Fixed in 1.4.13beta.
20070315: TCPIPTimeOut item in dicom.ini does not work. Timeout always 300 s. Fixed in 1.4.13beta.
20070316: The other cleanup runs when images are recieved. However, it works as follows: when a disk is full (< 30 mb), it cleans up until the given amount is reached. This may be considered a bug. Cleanup now starts at MAGDeviceThreshHold and runs until MAGDeviceThreshHold+1000. In 1.4.13beta.
20070316: After JPEG compression/decompression (0008,0008) CS has changed. This header is not reverted to the original even after decompresssion. This behaviour, I hope will be given a warning in the manual. This will bite anyone trying to save space ( Like those with multislice CT which can generated 3-4G of image data per study) Reconstruction on some 3D workstations on this modified files is not possible. The one we use, Leonardo from Siemens, will recongnize the changes. We can no longer use the files in the study for image postprocessing like Volume rendering and coronal sagittal reformations and others. This behaviour does not happen with NK compression. Does not seem to be a conquest problem.
20070331. When I try to anonymize a study and I don't enter a patient ID , Conquest comes back with an error message about Access violation, that can only go away if you stop the Conquest application through the task manager. Fixed in 1.4.13beta.
20070406. Henryk Kowalski: Linux version crashes in clib (most often in mallocpt) on 4 or more simultaneous inbound connections. Probably due to use of a non-thread safe clib function in conquest. Crashes happen faster with more inbound connections, and if logging is on but also occur when logging is off. Tested with various linux flavours - no difference. Windows version does not crash. To show crash location in linux run gdb dgate; run until crash; bt. Fixed in 1.4.13beta: log printing not thread safe.
20070410. Multiple simultaneous outgoing connections interfere: when CT is sending, and a move of a MRI is started to the same server, one of the two threads terminates prematurely. Fixed in 1.4.13beta.
20070410. Native mysql interferes with GUI shutdown in win2000. Unresolved.
20070507. Steini. I was using Native MySQL driver with Conquest 1.4.13Alpha and users started complaining that study descriptions did not match the study. And then later that referring physician was also "bullshit". (Well I did not check any other fields, this was bad enough)
Found out that using Native MySQL, Conquest somehow uses information from "previous" patient to fill in EMPTY study description fields and referring physician fields, when returning DICOM query requests. I changed to ODBC connection and it works normally now.
(MySQL ODBC 3.51 driver). Fixed in 1.4.13beta.
20070520: Nuccerz: I just regenerated the device using the "Regen Single Device" command. Normally, there's a message at the end of the operation which says "Regeneration Complete" or something to that effect. I haven't seen that message yet and it's been a few minutes since there was any change in the "Maintenance console". Meaning nothing is flashing by. When can I assume that it's done ? - Fixed in 1.4.13beta.
20070611: ajg: I have observed that when forwarding a series of studies to a different server using export with the ForwardAssociationLevel setting other than IMAGE, multiple images are sent in 1 association. However, when several studies are queued the second set of series or study are sent one at a time as if the association level reverted back to its default which is IMAGE. All subsequent studies are sent one image one association at a time. Fixed in final release 1.4.13.
20070701: several: native Mysql chokes when large number of images are dropped. Can be fixed through registry settings. Native mysql db creation will ask to set these in 1.4.13beta.
20070707: GUI crashes on XP when left for a long time in taskbar and then restored. Fixed in final 1.4.13
Marcel
Ted,
I'll be happy to include your code!
Marcel
