Internet mail => Network printer Internet mail => Offramp gateway (forward to G3Fax) Network scanner => Network printer Network scanner => Offramp gateway (forward to G3Fax) Network scanner => Internet mailMany of the participating groups included printers and scanners in their system. The items below were to be tested between the groups.
At the testing site, desks were in U-shape and seats were allocated to each group with 10Base-T hub and electric power. There were public telephone lines for tests of onramp/offramp gateway. Telephone line emulators were brought in by Cisco and WIDE Project. Telophone sets with regular G3FAX were available, too.
At the first, handouts consisting of lists were distributed. The lists included IP addresses and mail box's names allocated to each group as well as functions which could be provided by each. Individual interconnection tests were taken on these information. The table 1 is a summary of the identical contents.
Group | Function | Hardware etc. |
---|---|---|
A | SMTP-rcv,snd TIFF(S,F) DSN,MDN |
FAX=TELNUM@domain If busy(transient err),send to Mail-box |
B | SMTP-rcv,snd [LAN-POP] |
STN-T.30 B@domain |
C | SMTP-snd,POP-rcv / TIFF-FX | |
D | SMTP-snd,POP-rcv / TIFF(all) TIFF(all) |
|
E | SMTP-rcv,snd / TIFF(F) | @ |
F | SMTP-rcv,snd / TIFF(S,F) | |
G | SMTP-rcv,snd TIFF(S,F,J,C,L,M) SMTP-rcv,snd TIFF(S,F) SMTP-snd,POP-rcv TIFF(S,F) |
|
H | SMTP-snd,POP-rcv TIFF(S) |
IC FAX 3200 STN-T.30 |
I | SMTP-rcv,snd LAN/POP-rcv TIFF(S,F) |
TELNUM@domain |
J | SMTP-rcv,snd TIFF(F) POP-rcv,TIFF(FX) POP-rcv,TIFF(FX) | TELNUM@domain |
K | SMTP-rcv,snd |
[STN] <any>@domain |
L | SMTP-rcv,snd / POP-rcv | FAX=TELNUM@domain |
WIDE Project | SMTP-rcv,snd / POP-rcv TIFF(F) |
2-line STN available |
N | SMTP-snd / TIFF(S,F) | |
O | SMTP-rcv,snd / POP-rcv TIFF(S) | |
P | SMTP-snd,POP-rcv TIFF(S,F) |
PSTN-PPP |
Q | SMTP-rcv |
PSTN FAX=+TELNUM@domain |
We show the results of the connections between WIDE/IFAX and the other ones. Exchange of e-mails with 13 groups among 16 were confirmed. We had connection tests of onramp/offramp with 2 groups, and both tests succeeded.
A problem that we could not correctly handle e-mail messages from 2 groups happened during the experiment. The reason was that the senders made wrong e-mail headers. We had tests again after immediate discussion and fix. As a result, we could have correctly handled e-mail messages from one group. However we were not able to receive messages from the other group even though the reason and the way to fix were clear, because that group couldn't promptly manage.
We had as many as possible experiments with groups which had only a function of sending e-mail messages or receiving e-mail messages. It was confirmed that we could correctly exchange messages with all of these groups.
Onramp/offramp functions were tested by only WIDE Project and other 2 groups. When wholly considering FaxConnect 1, this is a future work for the next. FaxConnect 1 was over after the participating groups had an agreement that the next FaxConnect 2 would be held in May, 1999.
We had experiments to check interconnections with the other groups. This is to check that the others could correctly receive FAX data sent by us via SMTP and public telephone line, and we could correctly receive FAX data sent by the other Internet FAXes via SMTP and public telephone line.
The flow of the day is below.
9:00We prepared to connect with the others, having begun setting up the equipments and connecting. WIDE Project was allocated to just the corner of the U-shaped desks. A 10BASE-T hub was on the desk and a table tap under it. We had so many equipments that we connected our private table tap to that one, while we used the prepared hub with 16 ports.
We set up the main the Libretto in the following way.
After these, we configured a network on the Libretto. 3 IP addresses had been allocated per 1 group in advance. Because we had heard that extra petitions were available, we submitted an extra petition for more 2 addresses of our private terminals. Necessary information for network configuration, the participating groups etc. were distributed by handouts.
Then we checked behavior of WIDE/IFAX on the Libretto, to see if modules that send FAX data via SMTP would correctly work. Bugs were found here, we began to fix the program.
13:00Almost completed was fixing the bugs. It was capable of exchanging FAX data via SMTP.
15:00We began to negotiate with the others for practical connection tests. The following is the way to negotiate.
Results of the experiment via SMTP on this day were these.
Sent from WIDE/IFAX via SMTP | -- | 4 groups |
Received by WIDE/IFAX via SMTP | -- | 3 groups |
Any group did not have an experiment of onramp/offramp.
17:00The official experiment was over. We checked onramp/offramp behavior of WIDE/IFAX.
22:00We finished working.
The flow of the day is below.
9:00The experiments started. We had the experiments of SMTP and onramp/offramp. A way of connection tests of SMTP this day was same as the day before. Results of the experiment via SMTP on this day were these.
Sent from WIDE/IFAX via SMTP | -- | 11 groups |
Received by WIDE/IFAX via SMTP | -- | 11 groups |
WIDE/IFAX failed to receive from 2 groups of these. Suppose these 2 groups are X and Y respectively. The reasons were these.
No "To" field was there in headers of e-mail messages sent from the group X. WIDE/IFAX decides what to process to a message according to a mail header. Especially it judges whether the message is for itself or not depending on a value of the "To" field, so couldn't correctly process the messages from the group X.
In e-mail messages sent from the group Y, there was a problem at a value of the boundary option in the "Content-Type" field. At first, WIDE/IFAX could not correctly decode MIME because the value was too long. We discussed about this matter with the group Y. After this, what should have been a style of `\verb+ boundary="letters" +` was `\verb+ boundary=letters +`. Then WIDE/IFAX still failed to decode MIME.
The problem on the group X was solved after the discussion, but WIDE/IFAX could not correctly receive from the group Y at last.
11:00A checking sheet for verifying the implementation was conforming to each item in RFC 2305 was distributed.
12:00We demonstrated the onramp functions of WIDE/IFAX for a part of concerned people. This was that it converted a content received with G3 FAX to an e-mail message, sent this message to itself and output to a printer. The telephone line emulator used this time was brought in by WIDE Project.
13:00Delegates of the groups had a meeting.
15:00Delegates of the groups had a meeting again.
16:00Only groups capable of experiment had an experiment of onramp/offramp functions. This time a telephone line emulator brought in by Cisco was used.
Sent from WIDE/IFAX via G3 FAX | -- | 2 groups |
Received by WIDE/IFAX via G3 FAX | -- | 2 groups |
Closing.
We got mostly good results on connections via SMTP. Onramp/offramp functions were tested by only WIDE Project and other 2 groups. When wholly considering FaxConnect 1, this is a future work for the next. The groups agreed that the next FaxConnect 2 would be held in May, 1999.
This experiment made some problems clear. WIDE/IFAX tends to adopt many free softwares as a black box. This is a strong point of a system based on UNIX, and saved the development processes, too. But we got into trouble when an unclear problem happened in the black box. For instance, WIDE/IFAX depends on a program attached to a free software, HylaFax, when to make a TIFF format. We had believed that WIDE/IFAX sended formats in TIFF-F. But according to a result of a TIFF profile checker provided by Xerox after the experiment, the formats turned out to be both TIFF-F/S. Now we are strongly feeling that we should originally develop a TIFF analyzing program.
Although this is a small matter, we think over that we should have prepared a sample image in advance for tests. Of course any image had no problem on the experiment if it could be correctly sent. But we think an attractive sample image makes a group characteristic. WIDE Project had a contact with Tokyo to get a sample image comprising a picture of Mt. Fuji, but too late to use because it was made just before the end of the experiment. We want to deal with a matter like this in the next FaxConnect 2.
Overall, we could confirm correct connections with most of the groups, this is a good harvest, we think. But it was a pity that we couldn't have enough experiments on onramp/offramp. However, it is a fact indeed that WIDE/IFAX on which these functions are implemented has a big advantage. We are going to promote the development project for FaxConnect 2 in May, 1999.