UK Academic Community Directory Group Minutes of Meeting on 3rd October, 1991 Paul Barker Organisation: UCL Document Location: UCL _A_B_S_T_R_A_C_T Minutes of the Twelfth Meeting of the UK Academic Community Directory Group held at Brunel University on Thursday, 3rd October, 1991. October 10, 1991 UK Academic Community Directory Group Minutes of Meeting on 3rd October, 1991 Paul Barker Organisation: UCL Document Location: UCL Present: Shafiq Ahmed Nottingham Robert Ash Aberystwyth Adrian Barker Bloomsbury Paul Barker UCL (Secretary) Tim Beechey Stirling Colin Cooper Glasgow Jim Craigie JNT (Chair) John Cullen RAL Susan Feng Imperial Andrew Findlay Brunel Roger Gander OU Fred Greyer Queen Mary & Westfield Allan Hawdon Kings Robert Hogg Bradford Stuart Keir Sussex Jason Kitchen X-Tel Kevin Large UMCC Christina Lau Kent Qin Li Leeds John Linn Aberdeen Ravinder Lyall Southampton Damanjit Mahl Brunel Gareth Owen Aston Andy Powell Bath Colin Robbins X-Tel Graham Rule Edinburgh Alan Shepherd X-Tel Kel Shorey Strathclyde Alan Thew Liverpool Rodney Tillotson JNT Ray Walker DISC (guest) John Williams Aston - 2 - Apologies for absence from: Chris Bayliss Birmingham Steve Hardcastle-Kille UCL Julia Hill Heriot-Watt Caroline Leary Sussex Steve Titcombe UCL _1. _A_p_p_r_o_v_a_l _o_f _t_h_e _A_g_e_n_d_a The following agenda was agreed upon: 1. Introductions 2. Review of minutes of the meeting held on 13th June 1991. 3. Matters arising. 4. Quipu implementation status report 5. Quipu support status report 6. Brunel's user interface implementation status report 7. PARADISE DUA Development 8. X-Tel DUA 9. Directory pilot sites status reports 10. Report on the RARE Working Group 3 and RARE Directory Project 11. Report on PARADISE project 12. Report on IETF Directory Group 13. Registration of site's distinguished names in the UK Name Registration Authority operated by DISC. 14. Possible development of Directory Data Loading Engine 15. Proposed document on porting POD 16. Tutorial on ISODE daemons 17. Any other business - 3 - 18. Dates of future meetings 19. Workshop on operational problems _2. _A_p_p_r_o_v_a_l _o_f _m_i_n_u_t_e_s _o_f _p_r_e_v_i_o_u_s _m_e_e_t_i_n_g The minutes were approved without discussion - only about a third of those there had a copy of the minutes in front of them. The posting of documents had seemingly gone out too late for some people to receive them in time to collect them and bring them along to the meeting. _3. _M_a_t_t_e_r_s _A_r_i_s_i_n_g The PostScript version of the administrator's guide had not found its way into the document archives. Apparently there were problems producing a stand-alone version of the document in PostScript which would print successfully. It was felt that it would be useful to have electronic copies of the guide as the original was not of such print quality that it would photocopy well. A number of people had not received the guide. It appeared as though the mailing had not gone to all the right lists. An attempt would be made to make sure that all received the note. A discussion followed on the "chicken and egg" problem of promoting the service. Good quality data was required to be able to provide a good service. A good service was required to persuade data providers that they should make data available. This situation seems to blight a number of sites, and the impediment was as often as not a political one as much as technical reasons. Maybe referring to our efforts as a "pilot" rather than a "service" created the wrong impression? (This comment was made later in the meeting, but it seems to fit in well here.) It was not clear whether talking to administrators would be productive or not. In some cases where there had been substantial progress with the provision of directory services, it seemed as though the university authorities were not always well-informed about what was going on. Often when they did know what was happening, they got a bit worried. The official position, however, seems reasonably positive. Julia Hill had shown incredible cunning in actually trying to find out what the Registrars and Secretaries were saying at their meetings. (Her subtlety knowing no bounds, she discovered their views by actually talking to her university's secretary!) The wording of their resolutions was somewhat vague, but by no means totally negative. It seems as though collectively they are moderately brave, but individuals get nervous when isolated from other registrars and - 4 - secretaries, and take a less positive attitude. A report from Julia is reproduced as Appendix A. There was also a consensus that it would be a start if all computer centre directors felt that the directory was an essential service. This august body of men/women could then play a role in persuading their respective administrations to provide the necessary support. There was a need for more promotional material. The help-sheet produced by the PARADISE project to introduce the Directory and the de interface (this was circulated at the meeting) was a starting point as an introduction to directory services for ordinary users, and as a user guide for de. It did not appear to be entirely suitable in its current form. It needed to have a better general introduction to directory services. It also needed to be a little more practical and show how to make some typical queries. It was pointed out that the help-sheet was only as yet a first draft and it was hoped that people would make constructive comments as to its deficiencies as a user guide. These should be made fairly promptly as it was hoped to distribute the help-sheet at ESPRIT conference week in late November. Sites interested in copies of the revised sheet were asked to indicate how many copies they would like for local distribution. X-Tel were asked to produce a supplementary sheet with a practical focus, giving typical examples of usage. There was some discussion about which sites should run public access dua's. It was useful for all sites to do this as an initial lowest common denominator form of access until duas and interfaces had been ported onto PCs, VMS and other commonly used systems. Since individual sites would usually want to configure their interfaces to default to local access, no individual site was ideal as a JANET wide public access service. The PARADISE service did not as yet listen on a JANET address, and while it was possible to access this service across IXI, it was somewhat inefficient to do so due to the way that the calls were routed. It was decided that the JNT should run a JANET public access dua service, with a default country of "GB" configured. Initially de, sd and doog would be offered. Rutherford's mail responder had not yet been packaged and sent to X-Tel for distribution. Despite the fact that its packaging didn't appear to be an enormous amount of work, it had failed to command the day or so's attention that was thought to be required. The problems were essentially that the system came as a number of small pieces and that no thought had been - 5 - given to packaging when the system was being written. For example, at present it was only set up to look within the Rutherford name space. The other factor inhibiting the deployment of packaging effort was the lingering doubt over how useful the mail responder was to other people. Essentially it was useful to people who had electronic mail but rather impoverished networking in general - the situation of the IBM mainframe user is usually cited at this point. Rutherford would discuss the issue further with X-Tel. Similarly, Rutherford had not sent their DSA checker to X-Tel for assessment. Colin indicated that X-Tel were working on their own solution to the problem. It wasn't clear whether this would be made freely available to the UK academic community at the time of the meeting. Rodney Tillotson had produced a draft code of conduct, which he had circulated immediately prior to the meeting. Not surprisingly, it didn't please everybody! For each of the points on the proposed code, at least one person thought that that point was the weakest, or the most superfluous, ... Amongst the points raised was that it was impossible to say anything about compliance with usage of networks which may be on the other side of the world and which the user may not know they are even using. It was pointed out that ignorance of rules or laws did not excuse people for transgressing them. It was countered that weak rules, which were commonly flouted, served no purpose and were possibly counter productive. Someone noted that there was a need for a code of conduct for service providers as well as for users. The upshot of the discussion was that people should take the code back to their sites for discussion. Any code of conduct issues should be firmly within the ambit of directory services, as many aspects of a directory service code of conduct would almost certainly be covered in a more general code of conduct. Arising from these discussions, Andrew Findlay was to talk to the IUNC, to discuss the terms of usage of networks in general. It seemed likely that the current wording was already insufficient to cover all current aspects of network usage. The letter explaining sites' responsibilities with regard to hardware and software maintenance on the Directory machines had not yet been written (by Jim Craigie). This action is an "old friend" to many of us at the meetings! - 6 - Jim Craigie had spoken to SUN about the problem of future upgrades of the operating system only being available on CD-ROM, when the earlier machines in the pilot do not have CD players. There was still no final outcome to report. The JNT's position was that they wanted CD players for all the machines, or the tape distribution to be continued. Andrew Findlay had not provided details of what subset of ISODE was required to support the building and running of DUAs. Colin Robbins pointed out that a special version of ISODE had been produced for the PARADISE project which just built sufficient components of the ISODE release to build a DUA (the de interface) or a DSA. Graham Rule would look at what had been done and ascertain how useful a basis this was for interface porting in general. Colin Robbins said that ISODE had been ported to an ICL DRS6000 running SVR4. This port was of particular note as the machine had CONS! Colin was not sure whether the port was in the public domain at the time of the meeting, but said that it would be in ISODE-8.0 (no prospective release date at the time of writing). People were asked to note the existence of the portability database, maintained by X-Tel, which listed known ports of ISODE to various platforms. The secretary was informed that he was meant to be writing an evaluation of user interfaces - I shall try not to miss PARADISE project meetings in the future! It would still be beneficial if there was more feedback on user interfaces. The note on configuring DSAs to listen on IXI had been re-circulated and placed in the document archives. There were now 16 (out of 36) sites with an IXI address configured. Some of the problems with the probe producing suspicious figures had been remedied. My impression is that the figures were now thought to be a reasonably accurate reflection of reality. However, the usefulness of the results was still limited. The probe's raw data contained lots of useful information which would assist DSA managers to trace errors: in particular the times of failures. This information was lost in the overall summary. A request was made that information on probing failures should be sent to DSA managers. The secretary pointed out that it might not be worthwhile doing this at the moment if this turned out to involve a non-trivial amount of effort. There was a firm intention to rewrite the probe in C, and - 7 - thus further effort on the existing probe had to be limited. The developers would discuss the issue and report back to the group. The development of a VMS DUA had been discussed at a Network Users Group meeting. VMS users not surprisingly would like a DUA! The secretary mentioned that Peter Kay from New Zealand had recently reported to the isode mailing list that he had ported ISODE-7.0, including Quipu, to VMS. The secretary would forward this message to the group. And that was the first two and half hours of the meeting! _4. _Q_u_i_p_u _i_m_p_l_e_m_e_n_t_a_t_i_o_n _s_t_a_t_u_s _r_e_p_o_r_t ISODE-7.0, and thus Quipu-7.0, had been released in July. This was a major effort for X-Tel, who now control the ISODE source tree. There were a number of development areas where code had already been written or the designs were well advanced. Further access control had been added to searches to allow the imposition of limits on the number of results that could be returned. This feature is critical to wider scale deployment of the Directory. The replication mechanisms had been worked on, and it was now possible to do incremental update. Work on strong authentication was well advanced. Some more integration with the PP system had been done, in particular support of directory-based mail distribution lists. Work on a DSA console, somewhat similar to PP's MTA console, was well advanced. This allowed for the monitoring and control of DSAs - the control functions are broadly in line with those currently provided by dish's _d_s_a_c_o_n_t_r_o_l mechanism. Work had started on authorisation facilities. These offered access control, but where the focus was on access to resources rather than access to data. The scheme would allow the granting or denial of access to particular networks, for example, and would also allow the blocking of "rogue" DUAs and DSAs. Further thought had been given to the probe. Requests for additional facilities were still coming in! The probe will be rewritten in C soon - I'm being deliberately vague! - 8 - _5. _Q_u_i_p_u _s_u_p_p_o_r_t _s_t_a_t_u_s _r_e_p_o_r_t Colin Robbins introduced two new X-Tel employees, Jason Kitchen and Alan Shepherd, to the meeting. They would take over PP and Quipu support roles as their expertise increased. Colin reported that over half of the UK academic sites had upgraded to ISODE-7.0, and that the scripts had seemingly been helpful, although there had been some problems. A lot of sites had upgraded SUNOS on their directory machines. While many sites had good reasons for doing this (principally to keep their machines all running the same version of SUNOS), there was no Quipu-specific reason to upgrade SUNOS yet. A non Quipu-specific reason for upgrading is that SUN might well say that they can only offer full support on the latest versions of SUNOS. At least 3 sites had got themselves into a severe mess trying to upgrade SUNOS, and it had required considerable effort from X-Tel to get them "back on the road" again. It was not clear to X-Tel who supported the upgrading of SUNOS. SUN would offer help in this regard, but at the possible cost of losing the specific set-up used in the UK pilot. Why couldn't scripts be written to preserve site-specific tailoring, and reinstate it following the upgrade? The answer is that it would be very hard to do, to catch all the possibilities. It emerged at this point that the earlier machines in the pilot are configured slightly differently to the machines which have been pre- configured by X-Tel. The Rubicon would be crossed when Quipu required a SUNOS upgrade. This would be more X-Tel's problem, although it is not clear to the secretary at least where the boundaries of responsibility lie. Watch this space! In the interim the advice is that if you wish to upgrade SUNOS, preserve all your local tailoring, data, etc, and don't make too many mistakes. Would these be Directory Group minutes without a mention of CONS? Apparently not! Sites needed to run Sunnet-7.0 if they wanted to run CONS. If sites did upgrade to all the latest releases - SUNOS 4.1.1 and Sunnet-7.0, they would probably need a patch to prevent the machine panicking seemingly randomly. Glasgow had installed this patch and it seemed to fix the problem. Colin said that it was still too early to use CONS in the pilot. There were still a lot of technical problems with CONS, in particular with routing. Due to the nature of the problems, it would require a network - 9 - of test DSAs running over CONS to solve these problems, as it would adversely affect the reliability of the existing service if CONS addresses were added into the pilot DSAs. _6. _B_r_u_n_e_l _u_s_e_r _i_n_t_e_r_f_a_c_e _i_m_p_l_e_m_e_n_t_a_t_i_o_n _s_t_a_t_u_s _r_e_p_o_r_t Andrew Findlay gave a description of "xlu", for X LookUp, Brunel's latest X directory user interface. It was immensely configurable. It could be configured to provide a cut-down range of facilities, and offer simple look-up facilities. Alternatively, it could be configured such that a data administrator could use it for data maintenance. The focus of work was now on a version for PCs. The principal software platform was MS-Windows 3.0, although a VT100 mode would also be offered. The protocol stack provided by Edinburgh now appeared to be usable, although obviously there would be problems, given the nature and size of the software. It would be something of an understatement to say that a lot of work lay ahead for Brunel. _7. _P_A_R_A_D_I_S_E _D_U_A DE had been released as part of ISODE-7.0. However, some further work had been done subsequently on the interface to meet the PARADISE project's paymasters' acceptance criteria. The work was primarily bug-fixing and smoothing of rough edges. One useful feature had been added which allowed users to modify their terminal type (and some other variables) from within the interface. This would help users who discovered belatedly that their terminal emulation was working incorrectly. An upgrade package is available from the PARADISE helpdesk: email . _8. _X-_T_e_l'_s _D_U_A Colin Robbins announced that X-Tel had themselves produced a DUA, called XTDUA. It was a high functionality DUA, offering management facilities, and in particular offered support with setting up acls. A simpler browsing version was also available. The interface was an X-Tel product and would not be freely available. A demonstration version (for a SUN 4) was available for evaluation, and this had a lifetime of two months. The price was 300 pounds for the browsing-only version and 600 pounds for the version allowing modification and management facilities. This was the price per single copy. The - 10 - price would be reduced for volume sales - discuss this with X-Tel. _9. _R_e_p_o_r_t_s _f_r_o_m _D_i_r_e_c_t_o_r_y _P_i_l_o_t _s_i_t_e_s Thanks are due again to Rodney Tillotson for summarising the sites' reports. His summary is reproduced as Appendix B. There was a discussion about the usefulness of the reports. The secretary noted that there was too little real feedback from the reports about what the real problems were. Not many reports provided information on critical aspects, such as the adequacy of the software. The chair enquired as to whether people thought we should stop bothering with reports, but the feeling was that we should persist. The lack of feedback from ordinary sites, and the smallish group discussing the issues, was also noticeable to some extent at the meetings - there was very much a "top- table" and "others" dichotomy. Similarly the directory-group distribution list tended to be used for announcements and pronouncements, rather than for the discussion of issues. Noting that useful progress was usually made when there was a practical focus, such as at the Data Loading course, an experiment would be tried with the format of one of the coming meetings. Discussion would be curtailed to the morning session only, and there would be a tutorial/problems session in the afternoon. There were problems in that the host site had to be able to provide access to 10-15 terminals, but there were enough sites who could provide this facility. Several sites said that directory services were not yet seen as a key service and got a commensurate paucity of resources allocated to their development. This meant that sites really didn't have feedback to offer. There was general accord for the proposition that Whitemail would provide an enormous boost to the requirement for directory services. The importance of PC user agents was again emphasised. Aston reported that the Macintosh interface, produced by PSI, had been well received on their campus. This augured well for the day when more people had user agents which they could run from their desk-tops. The data management tools provided by Paul Barker and described at the Data Loading course should work with ISODE-7.0. A "problem" which had been discovered appeared to have been the result of trying to do things in too much of a hurry. The tools are available from - 11 - the UCL-CS info server in a file called dm.shar. _1_0. _R_e_p_o_r_t _o_n _R_A_R_E _W_G_3 Not given - lack of time. _1_1. _T_h_e _P_A_R_A_D_I_S_E _p_r_o_j_e_c_t The report was greatly curtailed due to time constraints. Sites were reminded to make comments on the help-sheet and to indicate if they wanted copies for local distribution. _1_2. _R_e_p_o_r_t _o_n _I_E_T_F _D_i_r_e_c_t_o_r_y _G_r_o_u_p Not given - lack of time. _1_3. _R_e_g_i_s_t_r_a_t_i_o_n _o_f _s_i_t_e_s' _d_i_s_t_i_n_g_u_i_s_h_e_d _n_a_m_e_s _i_n _t_h_e _U_K _N_a_m_e _R_e_g_i_s_t_r_a_t_i_o_n _A_u_t_h_o_r_i_t_y _o_p_e_r_a_t_e_d _b_y _D_I_S_C. The Directory Pilot has thus far worked with an ad hoc registration authority (Colin Robbins!) in the U.K. The small scale of the pilot has meant that we have not yet had to face up to the problems of resolving name clashes between competing organisations. The Directory Group had considered naming of organisations and had formulated some recommendations on naming. Jim Craigie had written a letter to university registrars setting out these recommendations, but this letter had not been sent. Following discussions with DISC, this letter had been revised, and would be sent to sites shortly. Ray Walker from DISC talked to the meeting about DISC's registration scheme, which would cover the registration of relative distinguished names in the namespace immediately below c=GB. The scheme was in some ways a step into the unknown. There was no clear anticipation on what the take-up on registrations might be. An attempt had been made to encourage registrations by pitching the registration cost as low as possible at 150 pounds. DISC would do a modest amount of policing of the scheme to try to avoid deliberately mis-leading registrations, but the low cost of the scheme implied that there would be limits as to how much checking it was feasible to do. Organisations could, however, register more than one name. Universities might consider registering both "University of XXX" and "XXX University", as this would prevent some dubious organisation from taking the other name form as their RDN. - 12 - The policy and some of the limitations of the scheme were interesting. DISC would allow a registration of the name "AA", for Automobile Association, while the advice of the Directory Group had been to choose long form names as RDNs and only to allow short forms as alternative values. The registration of the name "AA" might upset Alcoholics Anonymous", but basically the scheme was first come, first served, with only a modicum of moderation. There was a challenge procedure, whereby a challenge had to be made within one year of registration. There was obviously no experience of how this worked in practice yet. The scheme did not address the use of multi-valued RDNs, possibly using the locality attribute as a second naming component, as a way of disambiguating organisations. There were no guidelines as to where smaller organisations, not of national significance, should be registered in the DIT. The registration form also allowed for an object identifier request, and the registration of management domain names. Since you could do all these things for the price of a single registration, Ray Walker suggested that sites might as well go ahead and make these registrations, even if there was no current requirement for these other fields. Jim Craigie demurred at the suggestion of all sites registering their own management domains, foreseeing a slide into chaos compared with the cosy world of JANET! _1_4. _P_o_s_s_i_b_l_e _d_e_v_e_l_o_p_m_e_n_t _o_f _D_i_r_e_c_t_o_r_y _D_a_t_a _L_o_a_d_i_n_g _E_n_g_i_n_e Kel Shorey said that a data loading engine was an urgent requirement for the pilot. Too many sites were repeating work. Others pointed out that a part of the problem which he described would always remain site- specific; even if there was a data loading tool, there would always be a multiplicity of raw data formats, and that the current EDB or DM tool format was simple and about as general as one was going to get. The secretary said that there was a requirement for some data management tools but the development of these required proper funding. He knew only to well how bad the current tools were having written them! _1_5. _P_r_o_p_o_s_e_d _d_o_c_u_m_e_n_t _o_n _p_o_r_t_i_n_g _P_O_D Kel Shorey said that some effort should be put into porting POD onto platforms other than SUN. Colin Robbins said the software should run on a substantial number of UNIX platforms with little problem. Andrew Findlay said that it sounded like a configuration - 13 - problem and that the matter would be sorted out off- line. _1_6. _T_u_t_o_r_i_a_l _o_n _I_S_O_D_E _d_a_e_m_o_n_s Colin Robbins gave a tutorial on ISODE's daemons. The slides for his talk are available from the UCL-CS info-server: filename daemons.ps. _1_7. _A_n_y _o_t_h_e_r _b_u_s_i_n_e_s_s None. _1_8. _D_a_t_e (_a_n_d _p_l_a_c_e) _o_f _n_e_x_t _m_e_e_t_i_n_g_s. Thursday, 16th January 1992, in the Senate Room at the University of Birmingham. The meeting after that will be on Thursday, 9th April 1992, provisionally in Manchester. _1_9. _A_c_t_i_o_n_s _b_e_f_o_r_e _t_h_e _n_e_x_t _m_e_e_t_i_n_g. J Dyer To make available the PostScript of the administrator's guide so that it can be placed in the document archives. Rutherford To discuss the Directory mail responder RT To discuss Rutherford's DSA-checker with X-Tel RT To set up JANET public access DUAs. all To take the draft code of conduct back to their sites and to make constructive comments as to its improvement. JC Letter explaining sites' responsibilities with regard to hardware and software maintenance on the Directory machines. PB To forward Peter Kay's message on his port of ISODE to VMS to the directory group. X-Tel To provide an additional practical help sheet on de ?? all To make commments on the de help-sheet and to request, if required, copies of the sheet from the PARADISE helpdesk. - 14 - _A_p_p_e_n_d_i_x _A The following is an extract from the minutes of the steering committee of the Conference of Registrars and Secretaries held on 25/6/91. _S_T_A_F_F _A_N_D _S_T_U_D_E_N_T _D_A_T_A_B_A_S_E_S: _A_C_C_E_S_S _O_F _I_N_F_O_R_M_A_T_I_O_N - _X._5_0_0 _D_I_R_E_C_T_O_R_Y _S_E_R_V_I_C_E A brief description of the service had been circulated with the agenda. Steering Committee, while not opposed to the scheme, advised caution in respect of the inclusion of personal data on student and support staff in such a system. It remained highly desirable to screen initial enquiries to prevent unwanted contact. _A_G_R_E_E_D DAS That colleagues be advised that in general only information already available in published form (e.g. staff lists from the Calendar or internal telephone directory entries) be included in the X.500 Directory. (ii) That this matter be drawn to the attention of all members of CRS at the mid-term conference in September. Julia spoke to the Heriot-Watt Secretary following his attendance at the afore-mentioned conference. He made the following comments. 1. Some fears and concerns were raised. It was decided to allow only information currently available in mechanical systems to be used. 2. Once institutions used the directory and gained confidence in it, they would consider adding more info. 3. They would keep watching progress, and try to continue to control what went in. 4. People who wished to remain ex-directory would be allowed to do so. He said they were quite positive about the service, but wanted to see that people's civil rights were upheld - 15 - (wrt DPA etc). Julia's comment: I think that sounds quite positive also! Now we have to get the terminals on their desks so that they can use the service. This could be quite difficult. - 16 - _A_p_p_e_n_d_i_x _B Summary of Directory Pilot Site Reports, October 1991 Rodney Tillotson, Joint Network Team 1. Introduction This paper summarizes the Pilot Site reports produced for the JNT Directory Pilot meeting on 3rd October 1991. By Wednesday 2nd October, 28 of the 37 sites had sent a report. 2. Hardware and communications Four further sites (Bloomsbury, Glasgow, Nottingham and Oxford) have had disk trouble, in each case sorted out by Sun. Glasgow had other faults too. Leeds, RHBNC and Southampton are without X.25 access (Leeds and Southampton for want of cables). 3. Basic system software No problems are reported with any release of SunOS, nor with upgrades between them. Most sites are running SunOS 4.1 or later. A few sites are using 7.0 SunNet X.25. For basic use it seems reliable enough, though nobody has reported success in using CONS. 4. ISODE and DSA software Most sites are running ISODE 7.0 (the current release), having upgraded over the summer. The upgrade has in almost all cases been entirely straightforward; several sites report that 7.0 is more reliable than earlier releases, and X-Tel's support is well thought of. Faults reported (very few) are: CONS and shared libraries (Aston); Quipu crashes occasionally (Liverpool); Can't build static libraries (RAL). 5. DUA software All the DUAs from the pilot are in common use, and most sites have set up a public access DUA (usually dsc or de) as their local users' initial means of access. Some have mounted DUAs for local access only, which include pod and the osiwotsits. There are still calls for more or better DUAs, which are essential for wider - 17 - acceptance of the Directory. 6. Provision of data The data loading workshop was popular and apparently effective; several sites mentioned it, and are devising their own ways of liaising with administrators to obtain and install data. Julia Hill's glossy introduction is raising awareness, and this is generally helpful. Data in place: Bloomsbury, Liverpool, Nottingham, RAL, JNT (manual), KQC (political negotiations to extend), Stirling, Sussex (hoping to extend). Held up technically or for effort: Bath, Bradford, Brunel, Exeter, Heriot-Watt, Manchester, Southampton, Strathclyde, Warwick. Aston (work in hand), Oxford (little data). Held up politically: Aberdeen, Kent, OU, Surrey, Glasgow (opt-in), Imperial (still waiting). Still to start: Aberystwyth, Leeds, RHBNC. 7. Operational availability (%) 100 90+ 80+ 70+ 60+ 50+ 25+ >0 0 week to 14 Jun 0 2 12 6 2 2 2 2 4 21 Jun 1 13 6 1 2 1 1 2 6 28 Jun 0 9 5 3 3 2 3 3 5 5 Jul 2 13 2 2 3 3 4 1 3 12 Jul 1 13 5 1 2 0 6 3 2 19 Jul 0 16 3 0 3 1 5 1 4 26 Jul 0 2 10 4 6 3 3 1 3 2 Aug 9 10 4 1 3 0 4 2 2 9 Aug 0 16 7 3 2 2 1 3 1 16 Aug 0 20 2 4 2 2 2 2 1 23 Aug 1 15 4 3 2 1 3 3 3 30 Aug 1 18 4 1 2 0 1 4 3 6 Sep (results combined with the following week) 13 Sep 0 19 5 4 1 2 1 2 0 20 Sep 1 22 5 2 1 0 0 1 2 27 Sep 1 19 5 2 0 4 1 1 1 Availability distribution of DSAs as seen from JANET As in previous summaries, the interpretation of the probe results is uncertain; some sites feel that they are running all the time and don't understand why they are found to be unavailable. There are some weeks when everybody shows up badly and there is likely to be some explanation unrelated to the Directory. In normal weeks, median JANET - 18 - availability is nearly 90%. The presentation of probe results now includes a summary by network. IXI availability is consistently lower than JANET, although no explanation has yet emerged for that either.